[0001] This application claims the benefit of US Non-Provisional Application Serial No. 13/482,622 filed on May 29, 2012; US Non-Provisional Application Serial No. 13/482,249 filed on May 29, 2012; and US Non-Provisional Application Serial No. 13/482,638 filed on May 29, 2012, which are hereby expressly incorporated by reference in their entirety for all purposes.

BACKGROUND

[0002] This disclosure relates in general to serving requests for dynamic webpages and, but not by way of limitation, to accelerating serving requests for dynamic webpages with client- side scripting and requests for content files with embedded or referenced content objects.

[0003] Dynamic webpages allow programmers to tailor web pages for specific instances or viewers. Scripts may be executed on a client or server side, and the execution of the scripts may influence the appearance, content and/or functionality of a resulting webpage.

Programmers may utilize this flexibility to attempt to provide relevant content, pleasing displays, and useful features in a semi-customized manner. However, this flexibility also results in a lack of predictivity. Even if a same user requests a same webpage, the webpage displayed in a first instance may be drastically different as compared to the webpage displayed in a second instance.

SUMMARY

[0004] In one embodiment, the present disclosure provides a system and method for accelerating loading of dynamic webpages. A request for content file may be received from an end user system. The requested content file may be associated with a dynamic web page. The content file may be accessed and modified to include a reporting function. The modified content file is then transmitted to the requesting end user system. Subsequently, reports are received from the end user system that indicate characteristics of the dynamic web page. For example, the reports may identify parameters received by a client-side dynamic script or features of the web page as presented to the user (e.g., aesthetic features, content features, or functionality features).

[0005] The reports may be collected and aggregated, e.g., across users, and it may be determined which web-page features, content and/or functionalities are frequently utilized and how quickly subsequent to initial page loading. Based on this analysis, web-page features may be prioritized and acceleration techniques may be implemented to accelerate serving similar future requests. For example, content objects (e.g., images, HTML code portions, or Javascripts) may be prioritized based on which content objects were viewed or used first and/or fequently. High-priority objects may be pre-fetched and/or cached near the end user system (e.g., at an edge server), high-priority objects may be quickly or immediately transmitted to an end user system prior to receiving a request for the objects, codes may be modified to inline or embed high-priority content objects, portions of a content file may be rearranged and/or segmented, etc.

[0006] In some embodiments, a system is provided for serving dynamic content objects to end user systems. A request fulfiller receives a request for a webpage from an end-user system, retrieves a content file associated with the requested webpage, and transmits a modified content file to the end-user system. A content-file modifier generates the modified content file. The content- file modifier includes a dynamic-code detector that detects that the retrieved content file comprises or is associated with a dynamic code; and a reporting-code injector that injects a reporting code into the retrieved content file or an associated content file to produce the modified content file, the reporting code including instructions to report data identifying usage characteristics of one or more content objects. An object-usage analyzer analyzes the reported data. An object prioritizor that identifies a high-priority content object based at least in part on the analysis. An access improver that improves access of at least one end-user system to the high-priority content object. [0007] In some embodiments, a method is provided for serving content objects. A plurality of reports are received from a plurality of end-user systems, each report of the plurality of reports identifying one or more content objects presented or executed within a same dynamic webpage at a respective end-user system. Data from the plurality of reports is aggregated, wherein the aggregated data comprises an object-usage statistic. A content object of high priority is identified based at least in part on the object-usage statistic. An end-user system's access to the content object of high priority is improved, such that a time between a subsequent request for a Hypertext Transfer Protocol file associated with the dynamic webpage and an availability of the content object to the end-user system is reduced.

[0008] In some embodiments, a method is provided for responding to a request for a webpage. A request for a content file associated with a webpage is received from an end user system. The content file is retrieved. A detection that the webpage includes a dynamic webpage is made. A reporting function is injected into the content file or into another content file associated with the content file, the reporting function including instructions to collect and report data characterizing dynamic presentation or execution of content objects, the dynamic presentation of execution of the content objects being coordinated at the end-user system. The content file or another content file is transmitted to the end-user system.

[0009] In some embodiments, the present disclosure provides a system and method for delivering content objects over the Internet to an end user system. A request for a content file (e.g., for a HyperText Markup Language file defining a webpage) may be received from an end user system. The requested content file may be accessed and modified, such that content objects (e.g., JavaScript functions) embedded or referenced within the content file are replaced by stubs, and reporting instructions are injected into the code. When a stub is invoked, the associated content object is accessed and a report is generated identifying the stub access. Based on one or more of these reports, content objects associated with a content file are prioritized. Various techniques may be implemented to improve the speed at which high-priority content objects are available to users.

[0010] In some embodiments, code or links for high-priority content objects are re-injected into the code (replacing or negating functionality of the corresponding stub). Thus, these content objects may be accessed quickly, and rendering speed is improved by not transmitting low-priority content objects. Meanwhile, low-priority content objects remain accessible via the stubs. In some embodiments, high-priority content objects are collected and cached (e.g., at an edge server of an CDN), such that they are quickly retrievable.

[0011] In some embodiments, content objects are grouped into containers, and transmission of the container data is automatically initiated upon receiving a request for the associated content file. Thus, a webpage may be initially rendered, and an increasing number of embedded or referenced content objects may be transmitted to an end user system. Content objects may be ordered within a container, such that high-priority content objects are the firs to be transmitted. Prioritization of content objects may be applied to content objects of a variety of types. Thus, e.g., it is possible to determine a priority of a JavaScript relative to an image object. Content objects of various types may thus be included in a single container and transmitted in an order based on this inter-type prioritization.

[0012] In some embodiments of the invention, a system is provided for dynamically serving a content file with embedded or referenced content objects over the Internet to an end user system. A content object request function receives a request, from the end user system, for a webpage defined by the content file, the content file including a plurality of embedded or referenced content objects. A content-file modifier modifies the content file, the content- file modifier including a reporting-code injector that injects a reporting code into the content file, the reporting code including instructions to generate one or more reports with

information relating to the plurality of content objects. A report receiver receives one or more reports generated in response to implementation of the reporting code in the modified content file, the reports indicating for at least two of the plurality of content objects where the content object was rendered within the webpage. An object prioritizor prioritizes, in real time, the plurality of content objects, the prioritization depending on where the at least two of the plurality of content objects were rendered within the webpage. A delivery controller adapts, in real-time, one or more protocols for delivering at least some of the plurality of content objects to improve access to content objects assigned a high priority in the prioritization. [0013] In some embodiments, a system is provided for dynamically serving content files with embedded or referenced content objects over the Internet to an end user system. A content object request function receives a request, from the end user system, for one or more content files associated with a webpage, the one or more content files embedding or referencing a plurality of content objects. An as-rendered locator receives data from the end user system that indicates a location of at least some of the plurality of content objects within the webpage. An object prioritizor prioritizes the plurality of content objects, the

prioritization being based at least in part on the locations of the at least some of the plurality of content objects. A delivery controller adapts one or more protocols to improve access to content objects assigned a high priority, such that upon receiving a second request for the one or more content files associated with the webpage from a second end user system, the high- priority content objects are provided to the second end user system faster than they were to the end user system.

[0014] In some embodiments, a method is provided for dynamically serving content files with embedded or referenced content objects over the Internet to an end user system. A request, from the end user system, for one or more content files is received, the one or more content files embedding or referencing a plurality of content objects. Based on data received from the end user system, it is determined that a first content object of the plurality of content objects is made accessible to a user of the end user system before a second content object of the plurality of content objects is made accessible to the user of the end user system. The plurality of content objects are prioritized, such that the prioritization prioritizes the first content object over the second content object. One or more protocols for delivering at least some of the plurality of content objects are adapted to improve access to the first content object. [0015] In some embodiments of the invention, a system is provided for dynamically serving a content file with embedded or referenced content objects over the Internet to an end user system. A content object request function receives a request, from the end user system, for a webpage defined by the content file, the content file including a plurality of embedded or referenced content objects, at least one of the plurality of content objects including a JavaScript. A content-file modifier modifies the content file. The content-file modifier includes a reporting-code injector that injects a reporting code into the content file, the reporting code including instructions to generate one or more reports including information relating to invocation of stubs. The content-file modifier further includes a content-file parser that identifies each content object of the plurality of content objects embedded or referenced in the content file and a stub controller that replaces each of at least two of the identified content objects with a stub, wherein the stub requests the identified content object from a remote source upon invocation of the stub. A report receiver receives reports generated in response to implementation of the reporting code in the modified content file. An object usage analyzer dynamically analyzes the received reports to quantify usage of each of the stubs. An object prioritizor prioritizes, in real time, at least two content objects of the plurality of content objects based at least partly on the dynamic analysis of the received reports, the prioritization including an identification of at least one high-priority content object. An object injector modifies, in real-time, the content file such that the high-priority content object automatically loads upon rendering of the content file.

[0016] In some embodiments, a system is provided for dynamically serving content files with embedded or referenced content objects over the Internet to an end user system. A content object request function receives a request, from the end user system, for one or more content files, the one or more content files embedding or referencing a plurality of content objects, at least one of the plurality of content objects including a JavaScript. A content-file transmitter transmits a modified version of the one or more content file in response to the request, from the end user system, for the one or more content files. The modified version of the one or more content files includes a plurality of stubs, each stub being associated with a content object of the plurality of content objects. Each stub requests the associated content object from a remote source upon invocation of the stub. An object usage analyzer dynamically generates statistics identifying frequencies of invocations of the stubs. An object prioritizor assigns priorities to the content objects associated with the plurality of stubs in real time, the assigned priorities being based at least partly on the statistics dynamically generated by the object usage analyzer that identifies the frequencies of invocations of the stubs. [0017] In some embodiments, a method is provided for dynamically serving content files with embedded or referenced content objects over the Internet to an end user system. A request is received, from the end user system, for one or more content files, the one or more content files embedding or referencing a plurality of content objects. At least one of the plurality of content objects includes a JavaScript. A modified version of the one or more content file is transmitted in response to the request, from the end user system, for the one or more content files. The modified version of the one or more content files includes a plurality of stubs, each stub being associated with a content object of the plurality of content objects. Each stub requests the associated content object from a remote source upon invocation of the stub. Statistics are dynamically generated, the statistics identifying frequencies of invocations of the stubs. Priorities are assigned to the content objects associated with the plurality of stubs in real time, the assigned priorities being based at least partly on the statistics dynamically generated by the object usage analyzer that identifies the frequencies of invocations of the stubs. [0018] Further areas of applicability of the present disclosure will become apparent from the detailed description provided hereinafter. It should be understood that the detailed description and specific examples, while indicating various embodiments, are intended for purposes of illustration only and are not intended to necessarily limit the scope of the disclosure.

BRIEF DESCRIPTION OF THE DRAWINGS

[0019] The present disclosure is described in conjunction with the appended figures:

FIG. 1 depicts a block diagram of an embodiment of a content distribution system;

FIG. 2 depicts block diagrams of embodiments of a content delivery network (CDN) coupled to the Internet;

FIG. 3 depicts a block diagram of an embodiment of an edge server;

FIG. 4 illustrates a flowchart of an embodiment of a process for retrieving content objects;

FIG. 5 depicts a block diagram of an embodiment of a content- file modifier;

FIG. 6 illustrates a flowchart of an embodiment of a process for generating a modified content file;

FIG. 7 illustrates a flowchart of an embodiment of a process for reporting content-object use; FIG. 8 depicts a block diagram of an embodiment of an acceleration engine;

FIGS. 9a and 9b illustrate flowcharts of embodiments for prioritizing content objects;

FIG. 10 depicts a block diagram of an embodiment of content-modifying system; FIG. 11 illustrates a flowchart of an embodiment of a process for generating a modified content file;

FIG. 12 illustrates a flowchart of an embodiment of a process for accessing content;

FIG. 13 depicts a block diagram of an embodiment of components of the delivery controller; FIG. 14 illustrates a flowchart of an embodiment of a process for delivering content objects; FIG. 15 illustrates a flowchart of an embodiment of a process for determining a container property;

FIG. 16 depicts a block diagram of an embodiment of a computer system; and

FIG. 17 depicts a block diagram of an embodiment of a special-purpose computer.

[0020] In the appended figures, similar components and/or features may have the same reference label. Further, various components of the same type may be distinguished by following the reference label by a dash and a second label that distinguishes among the similar components. If only the first reference label is used in the specification, the description is applicable to any one of the similar components having the same first reference label irrespective of the second reference label.

DETAILED DESCRIPTION

[0021] The ensuing description provides preferred exemplary embodiment(s) only, and is not intended to limit the scope, applicability or configuration of the disclosure. Rather, the ensuing description of the preferred exemplary embodiment(s) will provide those skilled in the art with an enabling description for implementing a preferred exemplary embodiment of the disclosure. It should be understood that various changes may be made in the function and arrangement of elements without departing from the spirit and scope of the invention as set forth in the appended claims. [0022] In one embodiment, the present disclosure provides a system and method for accelerating loading of dynamic webpages. A request for content file may be received from an end user system. The requested content file may be associated with a dynamic web page. The content file may be accessed and modified to include a reporting function. The modified content file is then transmitted to the requesting end user system. Subsequently, reports are received from the end user system that indicate characteristics of the dynamic web page. For example, the reports may identify parameters received by a client-side dynamic script or features of the web page as presented to the user (e.g., aesthetic features, content features, or functionality features). [0023] The reports may be collected and aggregated, e.g., across users, and it may be determined which web-page features, content and/or functionalities are frequently utilized and how quickly subsequent to initial page loading. Based on this analysis, web-page features may be prioritized and acceleration techniques may be implemented to accelerate serving similar future requests. For example, content objects (e.g., images, HTML code portions, or Javascripts) may be prioritized based on which content objects were viewed or used first and/or fequently. High-priority objects may be pre-fetched and/or cached near the end user system (e.g., at an edge server), high-priority objects may be quickly or immediately transmitted to an end user system prior to receiving a request for the objects, codes may be modified to inline or embed high-priority content objects, portions of a content file may be rearranged and/or segmented, etc.

[0024] In one embodiment, the present disclosure provides a system and method for delivering content objects over the Internet to an end user system. A request for a content file (e.g., for a HyperText Markup Language file defining a webpage) may be received from an end user system. The requested content file may be accessed and modified, such that content objects (e.g., JavaScript functions) referenced or embedded within the content file are replaced by stubs, and reporting instructions are injected into the code. When a stub is invoked, the associated content object is accessed and a report is generated identifying the stub access. Based on one or more of these reports, content objects associated with a content file are prioritized. Various techniques may be implemented to improve the speed at which high-priority content objects are available to users.

[0025] In some embodiments, code or links for high-priority content objects are re-injected into the code (replacing or negating functionality of the corresponding stub). Thus, these content objects may be accessed quickly, and rendering speed is improved by not transmitting low-priority content objects. Meanwhile, low-priority content objects remain accessible via the stubs. In some embodiments, high-priority content objects are collected and cached (e.g., at an edge server of an CDN), such that they are quickly retrievable.

[0026] In some embodiments, content objects are grouped into containers, and transmission of the container data is automatically initiated upon receiving a request for the associated content file. Thus, a webpage may be initially rendered, and an increasing number of embedded or referenced content objects may be transmitted to an end user system. Content objects may be ordered within a container, such that high-priority content objects are the first to be transmitted. Prioritization of content objects may be applied to content objects of a variety of types. Thus, e.g., it is possible to determine a priority of a JavaScript relative to an image. Content objects of various types may thus be included in a single container and transmitted in an order based on this inter-type prioritization.

[0027] Referring first to FIG. 1, a block diagram of an embodiment of a content distribution system 100 is shown where a content originator 106 offloads delivery of the content objects to a content delivery network (CDN) 110. In this embodiment, the content distribution system 100 can dynamically and efficiently serve webpages and objects (e.g., JavaScript applets, images, videos, Flash animations, etc.) over the Internet 104 to end user systems 102 by determining characteristics of presented web pages and improving access to frequently used web page features. For example, a function (e.g., a document.write or document.writein function) or source code may be modified to include a reporting function that identifies what is displayed to a user, such that frequently presented or quickly presented web-page features may be identified. Frequently or quickly presented content objects may be cached, immediately provided, embedded, inlined, etc. Source code and/or style sheets may further be modified, e.g., to reduce uncertainties, decision points, etc. [0028] An end user may request a webpage, by e.g., entering a Uniform Resource Identifier or Uniform Resource Locator. The requested webpage may be associated with one or more files and/or objects, such as one or more source files (e.g., Hypertext Transfer files, files in a markup language, HTML files, dynamic HTML files, etc.). The one or more source files may include or reference (e.g., call) one or more scripts, such as one or more Javascripts. The scripts may be configured such that a result of the script is dynamically presenting a user with content, a functionality, display features, etc. For example, the scripts may include or call a document.write or document.writein function, and content may be dynamically selected from a set of available content, dynamically generated (e.g., based on user input), dynamically requested from a server, etc. [0029] Due to the dynamic capabilities of the script, a server responding to the request may have difficulty or be unable to identify features that may be used to generate the webpage. For example, it may be unaware that a script referenced or included in a source file will subsequently request an image. As described in greater detail below, some embodiments of the present invention relate to modifying files transmitted to an end user system, such that the end user system reports features about the dynamic generation of the webpage and/or features of the webpage as presented to the user. This data may then reveal frequently or quickly used content features. [0030] A content originator 106 produces and/or distributes content objects as the originator of content in a digital form for distribution with the Internet 104. Included in the content originator 106 are a content provider 108, a content site 116 and an origin server 112. The figure shows a single origin server 112, but it is to be understood embodiments could have multiple origin servers 112 that each can serve streams of the content object

redundantly. For example, the content originator 106 could have multiple origin servers 112 and assign any number of them to serve the content object.

[0031] Although, this figure only shows a single content originator 106 and a single CDN 110, there may be many of each in other embodiments. The content object is any content file or content stream and could include, for example, video, pictures, advertisements, applet, data, display characteristic (e.g., font, background, spacing), audio, software, and/or text. The content object could be live, delayed or stored. Thus, each content feature may be associated with one or more content files or file portions, such as an image file, audio file, video file, markup-language file (e.g., HTML file) or file portion, or style sheet file (e.g., CSS file) or file portion. Throughout the specification, references may be made to a content object, content and/or content feature, but it is to be understood that those terms could be generally used interchangeably wherever they may appear. Additional reference may be made to a content file. Generally, a content file may define a content object, content and/or content feature, though in some instances these terms may be used synonymously. For example, requesting a content object or requesting a content file associated with the content object may be used interchangeably.

[0032] Many content providers 108 use the CDN 110 to deliver the content objects over the Internet 104 to end users 128. When a content object is requested by an end user 128, the CDN 110 may retrieve the content object from the content provider 108. Alternatively, the content provider 108 may directly provide the content object to the CDN 110, i.e., in advance of the first request or in servicing the first request. In this embodiment, the content objects are provided to the CDN 110 through caching and/or pre-population algorithms and stored in one or more servers such that requests may be served from the CDN 110. The origin server 112 holds a copy of each content object for the content originator 106. Periodically, the contents of the origin server 112 may be reconciled with the CDNs 110 through a cache and/or pre- population algorithm. Some embodiments could populate the CDN 110 with content objects without having an accessible origin server such that the CDN serves as the origin server, a host or a mirror. [0033] The CDN 110 includes a number of points of presence (POPs) 120, which are geographically distributed through the content distribution system 100. Various embodiments may have any number of POPs 120 within the CDN 110 that are generally distributed in various locations around the Internet 104 to be proximate, in a network quality of service (QoS) sense, to end user systems 102. A wide area network (WAN) 114 or other backbone may couple the POPs 120 with each other and also couple the POPs 120 with other parts of the CDN 110. Other embodiments could couple POPs 120 together with the Internet optionally using encrypted tunneling.

[0034] When an end user 128 requests a content link through its respective end user system 102, the request for the content is passed either directly or indirectly via the Internet 104 to the content originator 106. The request for content, for example, could be an HTTP Get command sent to an IP address of the content originator 106 after a look-up that finds the IP address. The content originator 106 is the source or re-distributor of content objects. The content site 116 is accessed through a content site 116 in this embodiment by the end user system 102. In one embodiment, the content site 116 could be a web site where the content is viewable by a web browser. In other embodiments, the content site 116 could be accessible with application software other than a web browser. The content provider 108 can redirect content requests to any CDN 110 after they are made or can formulate the delivery path beforehand when the web page is formulated to point to the CDN 110. In any event, the request for content is handed over to the CDN 110 for fulfillment in this embodiment.

[0035] Once the request for content is passed to the CDN 110, the request is associated with a particular POP 120 within the CDN 110. A routing algorithm used to choose between different POPs 120 could be based upon efficiency, randomness, and/or proximity in Internet terms, defined by the fabric of the Internet and/or some other mechanism. The particular POP 120 then assigns or routes the request to an edge server. The particular POP 120 may retrieve the portion of the content object from the content provider 108. Alternatively, the content provider 108 may directly provide the content object to the CDN 110 and its associated POPs 120, i.e., in advance of the first request. In this embodiment, the content objects are provided to the CDN 110 and stored in one or more CDN servers such that the portion of the requested content may be served from the CDN 110. The origin server 112 holds one copy of each content object for the content originator 106. Periodically, the content of the origin server 112 may be reconciled with the CDN 110 through a cache, hosting and/or pre-population algorithm. [0036] An edge server serving the request may access the requested content - either by locally retrieving part or all of the content (e.g., from a local cache) or requesting it from another server. In some instances, the edge server determines a source for part or all of the requested content within the CDN 110 by querying other peer servers within or remote from the particular POP 120. This embodiment dynamically discovers peer servers, which have already cached or stored the requested content. The peer server that already holds the requested content could be an edge server or a server that doesn't service end user requests, for example, a relay server or ingest server. If part or all of the content cannot be found in the POP 120 originally receiving the request, neighboring POPs 120 could serve as the source in some cases, or the content could be sourced from the content originator 106.

[0037] Thus, a request from an end user system 102 for content may result in requests for content from one or more servers in the CDN 110. A CDN server (e.g., an edge server, peer servers, an origin server, etc.) may analyze requested content files (e.g., requested HTML files), modify the requested content files (e.g., to include reporting functions, embed or inline content objects, or modify scripts within or referenced within the content files), and transmit the modified requested content objects in response to the requests.

[0038] The end user system 102 processes the content for the end user 128 upon receipt of the content object. The end user system 102 could be a personal computer, media player, handheld computer Internet appliance, phone, IPTV set top, streaming radio or any other device that can receive and play content objects. In some embodiments, a number of end user systems 102 can be networked together sharing a single connection to the Internet 104.

Although this embodiment only shows a single content originator 106 and a single CDN 110, it is to be understood that there could be many of each in various embodiments.

[0039] In one embodiment, the content distribution system 100 can dynamically and efficiently serve objects (e.g., JavaScript scripts, images, videos, Flash animations, etc.) over the Internet 104 to end user systems 102 by determining usage patterns and differentially providing objects in view of the patterns. For example, objects generally used frequently or quickly after loading a webpage may be provided to users before objects generally used less frequently and/or later. As another example, objects used frequently may be automatically provided after a request for an associated webpage, whereas objects used infrequently may be provided only upon subsequent request for the object. In this way, frequently and/or quickly accessed objects are quickly provided to users and access to these objects is not slowed by transmitting objects accessed less frequently and/or accessed after a delay. [0040] In some embodiments, content distribution system 100 may further dynamically and efficiently serve webpage-code portions, style sheet portions, or style sheets (e.g., a

Cascading Style Sheet). For example, a webpage code (e.g., an HTML code) may be reorganized such that code producing webpage portions first used or viewed are presented immediately or towards a top of a code and other portions are embedded into the webpage, served subsequently and/or towards a bottom of a code. As another example, style sheets affecting a first-used or first- viewed portion of a webpage may be quickly available to a user system, and other style sheets may be subsequently provided. Thus, embodiments that relate to determining how quickly a content object is used or viewed and improving the probability that quickly used or viewed content objects are quickly available to user systems may be extended to HTML code portions, HTML codes, style sheet portions, and style sheets.

[0041] An end user may request a set of content object, e.g., by requesting a webpage associated with the set of content objects. For example, a user may request a file, such as a Hypertext Markup Language (HTML) file. The requested file may include multiple code portions, calls to multiple style sheets, and/or code that references or embeds a plurality of objects (e.g., including content objects, HTML code, etc.). For example, the file may include tags and code to embed an internal or external object (e.g., a JavaScript function) or libraries of objects into the file. A source of each of the content objects and/or the file may be on an edge server or any other server within the CDN. [0042] If an end user system retrieves all of the embedded and referenced objects before responding to the user's request, the response may be relatively slow and a substantial amount of bandwidth may be consumed. However, if an end user system does not retrieve all of the embedded and referenced objects before responding to the user's request, an object may be unavailable to display or run at an appropriate time. Thus, as described in greater detail below, embodiments of the present invention relate to differentially and/or selectively hastening the providing of embedded and referenced objects to an end user system based on predicted usage of the objects.

[0043] The content object is any content file or content stream and could include, for example, video, pictures, advertisements, script, data, audio, software, and/or text. The content object could be live, delayed or stored. Throughout the specification, references may be made to a content object, content, and/or content file, but it is to be understood that those terms could be generally used interchangeably wherever they may appear. [0044] Many content providers 108 use the CDN 110 to deliver the content objects over the Internet 104 to end users 128. When a content object is requested by an end user 128, the CDN 110 may retrieve the content object from the content provider 108. Alternatively, the content provider 108 may directly provide the content object to the CDN 110, i.e., in advance of the first request or in servicing the first request. In this embodiment, the content objects are provided to the CDN 110 through caching and/or pre-population algorithms and stored in one or more servers such that requests may be served from the CDN 110. The origin server 112 holds a copy of each content object for the content originator 106. Periodically, the contents of the origin server 112 may be reconciled with the CDNs 110 through a cache and/or pre- population algorithm. Some embodiments could populate the CDN 110 with content objects without having an accessible origin server such that the CDN serves as the origin server, a host or a mirror.

[0045] A CDN server (e.g., an edge server, peer servers, an origin server, etc.) may analyze requested content objects (e.g., requested HTML files), modify the requested content objects (e.g., to include reporting instructions, HTML stub wrappers, etc.), and transmit the modified requested content objects in response to the requests.

[0046] Referring next to FIG. 2, a block diagram of an embodiment of the CDN 110-1 is shown coupled to the Internet 104 with additional detail for one of the POPs 120. Each POP 120 may include a content request interface 220, a server selection function 224, a delivery interface 226, a number of CDN edge servers 230, and other servers (not shown). For simplicity, this embodiment only shows three POPs 120 within the CDN 110, however any number of POPs may exist in various embodiments.

[0047] As explained above in relation to FIG. 1, when a request for content is handed over to the CDN 110, the request is associated with a particular POP 120 within the CDN 110 using any number of algorithms. For example, the assignment could be round-robin, random, based upon loading of the edge server, distance between POP and end user, and/or other algorithms. The particular POP 120 receives the request through a content request interface 220 and distributes the request to the server selection function 224. The server selection function 224 assigns the request for content to an edge server 230 to stream the content object to the end user system 102. The server selection function 224 selects the edge server 230 from a group of edge servers 230 in the POP 120. A number of algorithms can be used to assign the request to an edge server 230. For example, the server selection function 224 may use routing algorithms, domain name service (DNS) resolution or an HTTP -redirect to direct a particular end user system 102 to a particular edge server 230. [0048] Various parameters may be taken into account for selection of the edge server 230. Examples of parameters influencing selection of the particular edge server may include content object characteristics, server assignment to a particular content provider, adequate quality of service (QoS), performance metrics, capabilities of the edge server 230, and/or routing efficiency between the edge server 230 and end user system 102. Embodiments could have any number of edge servers 230 within each POP 120 of the CDN 110.

[0049] When the request for content is assigned to the edge server 230, the edge server 230 may determine a source for part of all of the requested object (e.g., a content file and/or associated content objects). The content source(s) may be determined using one or more techniques. For example, a publishing point mapping content to one or more paths where the content can be found may be identified and/or a peer-discovery algorithm may be

implemented. In some embodiments, a content source may be determined by, e.g., querying a content registrar or distributed registrar. The content registrar may receive reports from edge servers in a plurality of POPs, and the distributed registrar may receive reports from edge servers in a particular POP. One or both registrars may be queried to identify where content is stored within the CDN. The content source(s) may include, e.g., the edge server 230 itself, a peer server or the content provider 108 or any other server with the content.

[0050] Some embodiments select one of the edge servers 230 to serve a particular content object for all requests in the POP 120. Where the edge server 230 receiving the request serves the content from a source external to its POP, like the content provider 108 or another POP 120, the edge server 230 would become the master for its POP 120 such that other edge servers 230 in the same POP would use the master edge server 230 found during peer discovery rather than getting the content externally. Other edge servers 230 receiving a request for the content object would find the master edge server 230 using peer discovery. [0051] Although this embodiment chooses a single edge server 230 other embodiments could assign the same content object request to multiple edge servers 230 on different occasions. To balance the load, for example, a variable number of edge servers 230 could be selected for a particular request such that they shared the delivery load. An edge server 230 without a particular content object could act as a relay to get the content object from a peer or neighboring edge server. The number of relay edge servers for a group of content object requests is configurable based on a number of factors, for example, stream name, customer name, service type, customer preference, number of edge servers, loading of edge servers, etc. [0052] Referring next to FIG. 3, a block diagram of an embodiment of an edge server 230 is shown. Edge server 230 includes a request fulfiller 305 that receives a request from an end user system 102 for one or more content objects. For example, an end user system 102 may be requesting content objects associated with a Uniform Resource Locator (URL) or

Hypertext Transfer Protocol.

[0053] The request fulfiller 305 may include a content-object identifier 310 that identifies content objects associated with the request. For example, a request for a webpage may in fact being requesting a plurality of content objects, such as HTML code for the webpage, included style sheets, images, Javascripts, etc. In some instances, it may not be possible to (at least initially) identify more than one content objects. For example, determining content objects associated with an HTML code may require first receiving and parsing the HTML code. However, in some embodiments, the associated content objects may be determined, e.g., based on past requests. For example, multiple previous requests may have been received for the same content object, and it may be determined that specific content objects were frequently or always requested shortly after the initial request. Thus, the subsequently requested content objects may be associated with the initial content object and identified by the content-object identifier 310 upon receiving a request for the initial content object.

[0054] The request fulfiller 305 may further include a local-storage detector 315 that detects whether one or more content objects are stored in a local storage or cache. For example, the local-storage detector 315 may determine whether a content- file database 320 includes one or more of the identified content objects. Local- storage detector 315 may include a look-up table or algorithm that indicates files stored in the database 320, or local- storage detector 315 may check whether specific content files are in the content- file database 320 (e.g., by requesting the content files). [0055] The request fulfiller 305 may further include a remote-storage requestor 325 that may request one or more content objects from a remote source. For example, the one or more content objects may be requested from another server (e.g., an edge server 230) in a same POP 120, another server (e.g., an edge server) in a different POP 120, an edge server 230, non-edge server, or an origin server 112. The remote-storage requestor 325 may determine where to transmit a request, e.g., by consulting a storage hierarchy, by consulting a storage look-up table, by consulting results of previous content-object remote requests, etc. In some instances, the remote-storage requestor 325 requests the one or more content objects after it has been determined (e.g., by the local-storage detector 315) that the one or more content objects are not stored locally. [0056] The edge server may further include an acceleration engine 330 that tailors content- object request strategies, caching strategies and/or transmission strategies in order to accelerate loading and/or execution of important one or more content objects (e.g., content objects desired to be quickly or frequently used on a webpage) on an end user system 102. [0057] The acceleration engine 330 may include a content- file modifier 335 that modifies one or more content files in order to determine use properties of a content object defined by the content file itself or referenced within (e.g., called by) the content file. The content file modified may include a requested content file (e.g., an HTTP file) or a content file associated with a requested content file (e.g., a script or function called by an HTTP file). For example, the modified content file may include a document. write or document. writeln function called by a script within or referenced by an HTTP file. The content-file modifier 335 may modify the content file to determine, e.g., with respect to one or more content objects (defined by the content file itself or referenced the content file): how the content objects are displayed to a user (e.g., near a top or bottom of a webpage), how frequently content objects are used, how quickly content objects are used, a delay associated with requesting and receiving the content object, etc. In some instances, the file modification allows the content-file modifier 335 to determine which additional content objects are associated with the content file (e.g., subsequently requested and rendered as a result of a script). The content- file modification may allow the content- file modifier 335 to determine properties associated with content objects not directly identified in the content file itself (e.g., dynamically identified by a script).

[0058] The acceleration engine 330 may include an object-usage analyzer 340 that analyzes properties related to when, how and how frequently specific content objects are used, as described further below. For example, the object-usage analyzer 340 may determine how frequently a content object was used or accessed and when it was used or accessed. The object-usage analyzer 340 may determine this information, e.g., based on reports received from end-user devices (e.g., as a result of the content-file modification) or by detecting subsequent requests from the same end user system 102.

[0059] Based on the analyzed properties of content objects, an object prioritizor 345 may identify priorities associated with one or more content objects, as described further below. Frequently and/or quickly used content objects may be highly prioritized. Other factors may further influence the priorities, such as transmission time, estimated content-object importance (e.g., prioritizing objects that allow webpage functionality, objects that allow proper webpage appearance, or objects that provide content to be presented), object size, etc. [0060] The acceleration engine 330 further includes an access improver 350 that improves access to one or more content objects. For example, the access improver 350 may speed up access of an end user system's access to high-priority content objects. High-priority content objects associated with a content file may be cached (e.g., at edge server 230, e.g., in the content- file database 320), pre-fetched, transmitted to an end user system 102 prior to receiving a request for the high-priority content objects, embedded or inlined into an associated content file, etc. In one instance, upon receiving a request for an initial content object (e.g., an HTML file), access improver 350 identifies high-priority content objects associated with the HTML file (e.g., images, scripts, other HTML files, etc.) and transmits the associated content objects to the end user system 102 prior to receiving a request for the content objects (e.g., the high-priority content objects may be transmitted substantially simultaneously with the initial content object.)

[0061] Although this embodiment shows a single edge server 230 having multiple components, it will be understood that in some instances, the components may be present across a plurality of edge servers 230 (e.g., one edge server having the request fulfiller 305, another edge server having the acceleration engine 330, and either or both have the content- file cache 320).

[0062] While FIG. 3 shows a variety of components in an edge server 230, in some instances other types of systems or devices include one or more of the depicted components. For example, a server other than an edge server in the content distribution system 100 (e.g., origin server(s) 112, one or more end user system(s) 102, a server coupled to a plurality of edge servers 230 (e.g., at a POP), by a server coupled to a plurality of POPs 120, an edge server 230, etc.) may include all or part of the acceleration engine 330. In some instance, the content distribution system 100 includes a plurality of acceleration engines 330, which may or may not be located in parallel locations (e.g., at a plurality of edge servers or at an edge server and at an origin server).

[0063] It will be understood that edge servers (within a POP and/or across POPS) may have similar, same or different components and functionality. For example, in some instances, some edge servers include an acceleration engine 330 and some do not. [0064] With reference to FIG. 4, a flow diagram of an embodiment of a process 400 for retrieving content objects is shown. The depicted portion of the process begins at step 405 where a request for a content file is received (e.g., by an edge server 230 and/or from an end user system 102). The requested content file may be associated with and/or at least partly define a webpage. The requested content file may comprise, e.g., an HTTP or HTML code and/or a script.

[0065] At block 410, one or more content files are identified transmissions (e.g., by the content-object identifier 310). The one or more content files may include the requested content file. In some instances, the one or more content files additionally or alternatively include other content files, such as scripts or other HTTP codes called by and/or otherwise associated with the requested content file. The other content files may be identified, e.g., based on a previous analysis of a similar or same requested content file and/or based on previously received use reports. For example, standard dynamic-webpage functions (e.g., document.write or document.writeln) may be associated with dynamic-webpage HTTP codes.

[0066] At block 415, at least one of the identified content files are accessed (e.g., by the local-storage detector 315 and/or remote-storage requestor 325). The content files may be accessed, e.g., from a local cache and/or from a remote cache or storage. At block 420, at least one of the content files are modified e.g., by the content-file modifier 335). The modification may cause an end user system 102 that executes the modified content file to report data back to an edge server 230. At block 425, the at least one modified content file is transmitted (e.g., by the request fulfiller 305 to the end user system 102) in response to the request. Additional content files (e.g., content files identified at block 410 but not modified) may also be transmitted.

[0067] At block 430, use data is received (e.g., by the object-usage analyzer 340), the use data indicating usage of one or more content objects. The one or more content objects may or may not include content objects associated with the content files identified at block 410 and/or associated with the modified content file. In some instances, the one or more content objects include content objects called or requested subsequent to rendering of a webpage associated with the modified content file. In some instances, the one or more content objects include content objects called or requested via execution of the modified content file.

[0068] At block 435, at least one high-priority content object is identified (e.g., by the object prioritizor 345) based at least in part on the use data. For example, frequently used or quickly used content objects may be highly prioritized. The high-priority content object may be identified based at least in part on the use data associated with the request received at block 405 and at least in part on one or more other similar requests (e.g., for similar content files, for content files associated with similar content objects, etc.). It will be understood that reference to "use" of a content object herein may indicate, e.g., presenting (e.g., a text-based or image content object), executing (e.g., a script), calling or including (e.g., a style sheet), etc.

[0069] At block 440, access to the at least one high-priority content object is improved (e.g., by the access improver 350). The access may be improved, e.g., by caching the high- priority content object at an edge server, transmitting the high-priority content object to an end user system 102 prior to a request for the content object, revising a content file to definitively include the high-priority content object (e.g., via embedding or inlining) rather than including it in a dynamic script, etc. [0070] Referring next to FIG. 5, a block diagram of an embodiment of the content- file modifier 335 is shown. A content file modified by content- file modifier 335 may be accessed, e.g., by intercepting the content file while the content file is being transmitted to another location. Thus, for example, the request fulfiller 305 may have requested the content file from an origin server, and the content- file modifier 335 may detect a response to the request and intercept the content file.

[0071] In some instances, a content file is accessed from the content-file database 320. The content-file database 320 may include original and/or modified content files. Original and/or modified content files may be associated with storage locations, naming conventions, associated identifiers (e.g., in the content file or in an index of content files), etc. in order to convey a modification status. The content-file database 320 may or may not be stored locally with respect to the content-file modifier 335. In some instances, one or more content-file databases 320 are located at each of a plurality of edge servers.

[0072] The content file may include a file written in a markup language (e.g., HTML), a scripting language and/or a style-sheet language (e.g., Cascading Style Sheets). The content file may include an HTTP file. The content file may include one or more embedded or inlined content objects or references (e.g., calls) to content objects. The content file may comprise a code, such as a code written in a markup language (e.g., Dynamic HTML or HTML). The content file may be modified prior to a compiling of the content file. The content file may comprise a single file or a set of files. The content file may comprise a webpage-defining file. [0073] The content-file modifier 335 includes a content-file parser 505. The content- file parser 505 may identify functions called (directly or indirectly) by the content file or content objects referenced or included within the content file. For example, the content-file parser 505 may identify inline content-object links (e.g., to images), embedded content objects (e.g., by scanning for tags), internal scripts, referenced functions, etc. In some instances, the content-file parse may identify portions of the content file (e.g., portions of an HTML code).

[0074] Content- file modifier 335 includes a dynamic-code detector 510 that determines whether a content file includes or uses a dynamic code, such that different content objects may be presented to different users. Dynamic codes may be structured such that it is difficult or impossible to determine, a priori, which content objects will or even might be used, e.g., during rendering of a webpage. The dynamic-code detector 510 may access the parsed content file and determine whether a portion of the content file indicates or suggests that the content file is or is associated with a dynamic code. For example, it may be determined, e.g., whether the content file includes or references a script (e.g., a script generally, a JavaScript, a client-side script, a script with particular features, a script that calls specific functions, a script that includes a document. write function or document. writeln, etc.).

[0075] The content-file modifier 335 includes a reporting-code injector 515, which identifies a target code and injects a reporting code into the content file. The reporting code may be added, e.g., to a dynamic code, scripting file, base code (e.g., that references or includes other content files but is not itself referenced or included by other content files), a secondary content file (e.g., referenced or included by a base content file), a script function, a traditional content file (e.g., a document.write or document.writeln function), a custom content file, a content file requested by an end-user system and/or a content file associated with a content file requested by an end-user system. For example, the reporting-code injector 515 may identify that a detected dynamic code will likely use the document.write or document.writelnfunction (e.g., based on its dynamic characterization, use of a JavaScript generally, use of a particular type of script, etc.) or may specifically detect a call to the document.write or document.writelnfunction. The document.write or document.writeln function may then be identified as the target code.

[0076] The reporting-code injector 515 may then inject a reporting code in the target code. The reporting code may include instructions to generate data indicating, e.g., when the target code was called, when a portion of the target code was executed, one or more parameters sent to the target code, an output of the target code and/or one or more content objects identified or presented by the target code. For example, a reporting code injected into a document.write or document.writelnfunction may instruct the function to generate data noting which parameters were input into the function and when the function was called. This information may indicate any identity of content objects presented to a user and when the presentations occurred. [0077] The reporting code may further include instructions to format the data and/or transmit the data (e.g., to the edge server 230). The reporting code may further include transmission conditions (e.g., that new parameters were received or that at least a specific amount of time has passed since a previous transmission). Thus, for example, a reporting code injected into a document. write or document. writeln function may instruct the function to generate a report identifying parameters received in the last 2 minutes and transmit the report to the edge server 230 (or to another destination, such as another edge server 230).

[0078] The reporting code may be injected, e.g., towards a beginning of the content file, near calls to functions, near references to content objects, near embedded objects, etc. In some instances, multiple reporting codes are injected (e.g., in multiple content files or a same content file). In some instances, the reporting code includes multiple conditions or instruction dependencies. For example, instructions may indicate that a first type of data is to be collected when a script causes text to be displayed (e.g., the data identifying that text generally is being displayed) and a second type of data is to be collected when a script presents a content object (e.g., the data identifying the content object).

[0079] In some instances, the reporting code includes instructions to identify content objects currently loaded, displayed, and/or being executed. The reporting code may include instructions to further identify where (e.g., along a y-axis), within a webpage, the content objects are displayed. For example, some content objects may be displayed within a top portion of the page that is immediately viewable to a user, and other content objects may be displayed within a bottom portion of the page that is only viewable to the user upon scrolling down the page. By identifying web-page placement of content objects and tracking scrolling inputs, a determination may be made as to whether specific content objects of the webpage are currently within a field of view. [0080] The modified content file may be stored, e.g., in a content- file database (such as content-file database 320) at an end user system, at a POP, at an edge server, at an origin server, etc. The modified content file may be stored in a local or remote location. With respect to the content- file modifier 335. In some instances, multiple versions of the modified content file are saved. For example, the modified content file may be generated at or near the origin server, and copies of the modified content file may then be stored at a plurality of (or each) POP within a CDN or at a plurality of (or each) edge server within a CDN. Copies may then be revised independently from each other and/or in a global manner. [0081] The modified content file may be transmitted, e.g., to an end user system. For example, the modified content file may be transmitted to an end user system that requested the content file or that requested another content file associated with the modified content file. The modified content file may include a same or different name as compared to a name of an unmodified version of the content file.

[0082] With reference to FIG. 6, a flow diagram of an embodiment of a process 600 for generating a modified content file is shown. The depicted portion of the process begins in step 605 where the content-file modifier 335 accesses one or more content files (e.g., received by the from a remote source or accessed from a content-file database 320). At block 610, the content file is parsed (e.g., by content- file parser 505). The content file may be parsed, e.g., to identify internal scripts, links, function calls, file inclusions, etc. In some instances, multiple content files are parsed. For example, after an initial parsing, it may be determined that the initial content file includes a second content file. Thus, the second content file may be accessed and parsed. [0083] A determination is made, at block 615, as to whether an accessed content file includes or is associated with a dynamic code (e.g., by the dynamic-code detector 510). If the content file does not include and/or is not associated with a dynamic code, then then the process continues to block 630, and the modified file is transmitted (e.g., to an edge server, end user system, etc.). [0084] If the content file includes and/or is associated with a dynamic code, then processing continues to block 620, at which a target code interacting with dynamically requested content objects is identified (e.g., by reporting-code injector 515). The identified target code may include a code that, e.g., dynamically identifies, requests (e.g., from the content delivery network 110), formats, or presents (e.g., to an end user system 102) a content object. The target code may be known or estimated to be included or called by the dynamic- code content file. The target code may be included in or may be, e.g., a function or a script.

[0085] At block 625, a reporting code is injected into the target code (e.g., by reporting- code injector 515). The reporting code may be, e.g., injected at a beginning, middle or end of the code. The reporting code may be textually included in the code or a call or external inclusion may be added from the target code to a discrete reporting code. At block 630, the modified target code is transmitted (e.g., to an end-user system). The modified target code may be included in a requested content file and/or may be transmitted substantially simultaneously to transmission of a requested content file. [0086] With reference to FIG. 7, a flow diagram of an embodiment of a process 700 for reporting content-object use is shown. The depicted portion of the process begins in step 705 where a content file (e.g., a webpage) is requested. For example, an user 128 may enter a uniform resource locator (URL), click on a hypertext link, etc. The request may be transmitted, e.g., from an end user system 102 associated with the user 128 over a network (e.g., the Internet 104) to a POP 120. The request may be assigned to an edge server 230 at the POP.

[0087] In response to the request, at block 710, a modified content file may be received by an end user system 102. For example, an edge server 230 may receive and modify the requested content file in accordance with process 600 depicted in FIG. 6 and transmit the file to the end user system 102, or an origin server may access and modify the requested content file in accordance with process 600 depicted in FIG. 6 and transmit the file to the end user system 102 via an edge server. In some instances, the end user system 102 receives an unmodified content file and itself modifies the file, e.g., in accordance with process 600 depicted in FIG. 6.

[0088] At block 715, a dynamic script associated with the modified content file is executed. For example, the modified content file many include a dynamic script or may be called by a dynamic script. Execution of the dynamic script may result in dynamically determined content objects to be presented. Thus, at block 720, one or more content objects are dynamically requested or rendered. For example, the dynamic script may process inputs (e.g., from the user 128 or from external sources) and identify or select one or more content objects to be presented. In instances in which the identified content objects are not locally available, the content objects may be requested (e.g., from the content delivery network 110). In some instances, the content objects are locally available (e.g., in a cache) or are locally generated (e.g., generating an image based on inputs). The dynamically determined content objects may then be rendered.

[0089] At block 725, the request and/or rendering of the content objects may be reported (e.g., to an edge server 230). For example, report data may indicate which content objects were rendered, when they were rendered and/or where (within a webpage) they were presented. The reporting may be a result of a reporting-code injected into the modified content file. Process 700 may involve repetition of blocks 720-725, such that content objects may be repeatedly dynamically requested and/or rendered and report data identify the dynamic use of the content objects may be repeatedly collected and transmitted. [0090] Referring next to FIG. 8, a block diagram of an embodiment of an acceleration engine 330 is shown. In the depicted embodiment, the acceleration engine 330 does not include a content-file modifier 335. The content-file modifier 335 may, e.g., be present in another acceleration engine. In some embodiments, the acceleration engine 330 as depicted in FIG. 8 further includes the content-file modifier 335. The acceleration engine 330 may be present in and/or coupled to any of a variety of components in the content distribution system 100. For example, in some instances, the acceleration engine 330 may be implemented by each of a plurality of edge servers 230 (e.g., that respond to requests from users 128) at each of a plurality of POPs 120. Other implementations are possible. For example, the acceleration engine 330 may be implemented by the origin server(s) 112, by one or more end user system(s) 102, by a server coupled to a plurality of edge servers 230 (e.g., at a POP), by a server coupled to a plurality of POPs 120, etc. In some instance, the content distribution system 100 includes a plurality of acceleration engines 330, which may or may not be located in parallel locations (e.g., at a plurality of edge servers or at an edge server and at an origin server).

[0091] The acceleration engine 330 includes a report receiver 805, which receives reporting data identifying usage patterns as they pertain to one or more content files. The reports may indicate, e.g., which content objects were accessed, properties of accessed content objects (e.g., the type of object), when content objects were accessed, how content objects were accessed and/or how content objects were presented. The reports may include, e.g., one or more time stamps, such as a time stamp associated with: a request for a content file, with receipt of the content file, display of a webpage associated with the content file to a user, etc. The time stamp may include an absolute time or a relative time (e.g., relative to a time of a request for the content file). The reports may include client-side information that, e.g., is would not otherwise be available to the content delivery network 110 without the reporting data. For example, the reports may include information regarding client-side generated content objects, appearance of dynamic webpages, times at which content objects (e.g., previously transmitted to the end user system 102) are presented, etc.

[0092] The reports may include details about an end-user system, such as a browser used to display content in the content file, an operating system, a type of device, etc. Reports may be received from one or more users 128 and/or end user systems 102. In some embodiments, end user systems 102 that were served a content file (or a modified version thereof) from an edge server or from an origin server (e.g., via an edge server) send reports to the edge server 230 or to the origin server 112. Reports may also be received from parallel or higher level network components. For example, a report receiver 805 at a first edge server 230-1 may receive reports from a second edge server 230-2 (which may have received the reports from an end user system 102).

[0093] In some instances, reports may be generated by and received (directly or indirectly) from end user systems 102. In some instances, reports are additionally or alternatively generated by and received from other servers. For example, a skeleton residing on a server remote from an end user system may generate a report after being called by a corresponding stub or after responding to a call from a corresponding stub.

[0094] The acceleration engine 330 includes an object-usage analyzer 340, which analyzes the reports received by the report receiver 805. The analysis may be performed, e.g., for one, more or each content object: identified in at least one report associated with a requested content file or webpage associated with the report, transmitted to the end-user system in response to a request for a report-related webpage, that may be possibly displayed on a report-related webpage, etc. The object usage analyzer 340 may determine whether, how frequently, when and/or how the content object was accessed.

[0095] An access detector 810 may determine whether a content object associated with a content file was accessed generally or within a particular time period. Access to the content object may be in the received reports or may be inferred based on data in the reports. For example, the access detector 810 may detect access to a content object when a report indicates that a dynamic script identified, generated or selected the content object or when a presentation or processing function received the content object.

[0096] A usage time-stamp identifier 815 may determine temporal information related to access of the content object. For example, the usage time-stamp identifier 815 may identify an absolute time of access of a content object or a relative time of access of a content object (e.g., relative to a display of a corresponding webpage, request for the content file, etc.). The usage time-stamp identifier 815 may identify an order of access to content objects. For example, a user may have accessed Content Object #1 first, Content Object #5 second, and Content Object #10 third.

[0097] Outputs from the access detector 810 and/or from the usage time-stamp identifier 815 may be aggregated across reports by the data aggregator 820. Data aggregator 820 may determine, e.g., outputs corresponding to a same or similar content file (e.g., being generated based on data from one or more reports) and aggregate the data. Thus, for example, the aggregated data may include data received from reports generated from a plurality of end user systems 102. The data may be aggregated, e.g., in a cumulative manner or in a time-binned manner.

[0098] One or more object usage statistics, stored in an object-usage-statistics database 825, may be generated based on the aggregated data. The object usage statistics may or may not include variables specific to a same or similar content file (e.g., URL). The object usage statistics may be specific to content objects. The object usage statistics may indicate, e.g., a frequency at which a content object was accessed, information pertaining to a relative time at which the content object was accessed (e.g., a mean time of access with respect to a time of a request for an associated content file), dependencies on locality or end-user-system variables (e.g., browsers, edge server locations, etc.), etc.

[0099] Based at least partly on the analysis performed by the object usage analyzer 340, an object prioritizor 345 then prioritizes content objects. The prioritization may include ranking at least two of the content objects, assigning the content objects to groups based on priorities, etc. [0100] In one instance, the prioritization comprises identifying one or more rarely- requested content objects. The rarely-requested content objects may comprise content objects associated with statistics identifying a low count of accesses and/or a low frequency of accesses. Thus, a count or frequency of access may be compared to an absolute threshold (e.g., identifying content objects requested fewer than 50 times or fewer than 50 times per month). A count or frequency of access may be compared to a relative threshold (e.g., identifying content objects for which access counts were less than 5% as compared to a maximum access count; identifying content objects for which access counts' contribution to a content-file's total access to count to all content objects was less than 1%, identifying content objects that were not amongst the top 20 most requested content objects, etc.). In one instance, the prioritization comprises determining which content objects are accessed quickly and which content objects are accessed after a delay.

[0101] Generally, frequently requested content objects may be prioritized over rarely requested content objects, and quickly accessed content objects may be prioritized over content objects accessed after a delay. However, other factors may further influence prioritization, such as: size of content objects, bandwidth consumption of transmitting the content object, estimated download time associated with the content object, use of a content object in multiple content files, popularity of an associated content file, etc. [0102] In some instances, the prioritization depends at least partly on an actual or predicted location of one or more content objects within a webpage as rendered. For example, a reporting code in a modified content file may instruct an end user system to collect (in a raw or processed form) information about a location of one or more content objects (e.g., including those defined inside iframes) with respect to a rendered, e.g., webpage. The information may be transmitted in a raw or processed form to an as-rendered locator 830. The information may itself comprise location estimates or the as-rendered locator 830 may estimate locations based on the data. Objects may be prioritized at least in part based upon the estimated location (e.g., prioritizing objects based on their position with respect of the Y- axis, such that top-positioned objects are assigned relatively high priorities).

[0103] The prioritization may be performed globally or for a particular set of data. For example, the prioritization may be specific to a content file (e.g., ranking content objects embedded in the content file), to a user characteristic (e.g., browser or user device), to a type of content object (e.g., Javascripts or images), etc. A location of the acceleration engine 330 may further influence a specificity of the prioritization. For example, an acceleration engine 330 located at an edge server and/or only receiving reports associated with one edge server may identify object prioritizations specific to the edge server. Thus, in some embodiments, acceleration engines 330 may be included within a content distribution system 100, and multiple object prioritizations may be determined. In some instances, prioritization schemes vary across different prioritizations. For example, in some embodiments, prioritizations of images depend upon their as-rendered locations, while prioritizations of Javascripts do not.

[0104] The acceleration engine 330 may include an access improver 350 to improve access to content objects assigned high priorities by the object prioritizor 345. In some instances, the access improver 350 coordinates automatic loading of the content objects assigned a high priority and/or identifies an order for automatically loading of the content objects based upon priorities. Thus, if a request for a content file is received after a prioritization performed by the object prioritizor 345, content objects assigned high priorities (e.g., and predicted to be frequently used and/or used quickly) may be quickly or immediately available to a user following rendering, e.g., of a webpage. This may be accomplished by, e.g., revising a content file (e.g., to include a content object code), automatically transmitting another file and/or high-priority content objects in response to subsequent content-file requests, coordinating automatic transmission of content objects via virtual containers to prioritize the high-priority content objects, and/or caching high-priority content objects near end user systems 102 (e.g., at one, more or all edge servers 230). For example, code may be inserted into the content file to automatically load a high-priority content object, and an associated stub may be removed. In some instances, stubs are not removed. A stub may, e.g., include conditional instructions, such that a content object is only requested if it has not already been loaded. Therefore, even if the stub remains in the file, it will not substantially affect the functionality of the code, as it will not request objects already available locally.

[0105] In some instances, content files (e.g., stored in the content-file database 320) are modified. For example, an object injector 840 may inject content objects (e.g., high-priority content objects) and/or code associated with the content objects (e.g., load instructions, direct links, local links, etc.) into the modified content files. Thus, accessing the high-priority content objects may be more direct and result in faster access. In some instances, at least a portion of a dynamic code or a reference to a dynamic code is removed.

[0106] As another example, a content file may be rearranged such that HTML-code content objects are ordered and/or segmented according to a priority. HTML code portions producing webpage portions viewed first may be associated with a high priority and pulled towards a top of an HTML code. HTML code portions producing infrequently viewed webpage portions may be associated with a low priority, and pushed towards a bottom of an HTML code or segmented into a separate HTML file called by the initial HTML code.

[0107] In some instances, high priority objects are collected by an object collector 845. The collected content objects may be stored or cached, e.g., in a content-object database 850. The content-object database 850 may be located, e.g., at an edge server. Thus, requests for a content file may result in requests sent to fewer, e.g., origin servers, and the speed required to respond a request for the content object may be reduced.

[0108] A delivery controller 855 may further generate and/or adapt one or more delivery protocols (e.g., stored in a delivery-protocol database 860) to improve access to high-priority content objects. For example, high-priority content objects may be automatically transmitted to an end user system 102 upon receiving a request for an associated content file. The automatically transmitted content objects may further be divided into multiple groups (to allow for multiple simultaneous transmissions) and/or ordered (e.g., within a group) to improve the quick availability of high-priority content objects. For example, FIGs. 13-14 and the related text describe embodiments related to grouping content objects into containers.

[0109] With reference to FIG. 9a, a flow diagram of an embodiment of a process 900a for prioritizing content objects is shown. The depicted portion of the process begins in step 905 where an object-usage report is received by the report receiver 805. The object-usage report may include data identifying, e.g., that an object was identified by a script, that an object was received by a function, that an object was presented or requested, or that an object was otherwise accessed or interacted with. The object-usage report may include data identifying access to one or more content objects, associated with one or more files. The received report may include information about when the one or more content objects were accessed.

[0110] The access detector 810 detects, at block 910, access of one or more content objects based on data in the received report. The usage time-stamp identifier 815 identifies, at block 915, temporal features related to access of one or more content objects. Blocks 905-915 may be repeated to accumulate data pertaining to a content file. Subsequently or simultaneously, at block 920, the data aggregator 820 aggregates the object-usage data (e.g., including detected access and/or identified temporal features) and generates usage statistics.

[0111] Based at least in part on the generated usage statistics, the object prioritizor 345 prioritizes objects (e.g., associated with a particular content file or a group of content files, such as a dynamic webpage) at block 925. At block 930, an access improver 350 improves (e.g., hastens) access to high-priority objects. The process 900a may be repeated, such that objects are dynamically and frequently prioritized. Further the process 900a may be performed in real-time or near real-time. Thus, no separate training period is necessary, and objects may be prioritized and access to objects adjusted nearly immediately after a content file is available over a network. [0112] With reference to FIG, 9b, a flow diagram of an embodiment of a process 900b for prioritizing content objects is shown. Many blocks parallel those in FIG. 9a, and pertinent related disclosures are contemplated for this embodiment as well. FIG. 9b shows examples of how access to high-priority objects may be improved at block 930 in FIG. 9a. [0113] At block 950, a content file may be revised. For example, the object injector 840 may inject a high-priority content object and/or code associated with a high-priority content object (e.g., a JavaScript function, a direct and/or local link, etc.) into a content file and/or may remove at least some of a dynamic eode in or associated with the content file. As another example, a content file may be reorganized such that high-priority portions (e.g., file-portion content objects) or high-priority calls or towards a top of the content file. As another example, a. content file may be segmented such that low-priority content objects (e.g., HTML eode or an embedded object) are reorganized into a separate code that is called by the revised content file. [0114] At block 955, the object collector 845 may collect high-priority objects. The collected high-priority objects may be cached, at block 765, e.g., at a local server or edge server in a content-object database. As described in further detail below, the collected objects may be stored in or associated with virtual containers. At block 970, the deliver}' controller 855 may control deliver}' of the content fi le and of content objects. For example, the delivery controller 855 may determine which content objects to transmit with the content file and/or to transmit automatically upon receiving a request for the content file. The delivery controller 855 may further determine an order of transmission of content objects, a number of simultaneous downloads (of content objects) to initiate, compiling conditions, etc,

[0115] Thus, some embodiments herein provide accelerating measures for dynamic code. Dynamic codes are traditionally associated with unpredictable content-object usage and may result in repeated content-object requests (e.g., as a result of multiple iterations or nested script structures). Serving each request individually is time consuming, delays users' access to content, and wastes processing power on the server and client side.

[0116] Meanwhile, embodiments herein allow for a server-side device to identify client- side usage and/or to determine how a dynamic code is operating. Thus, the outputs of the dynamic code become more predictable. Based on this information, frequently used content objects may be made quickly available for use. For example, the content objects may be cached near an end user system 102 or transmitted to an end user system 102 prior to receiving a request. Processing may be saved on the client side, as an end user system 102 may identify that a particular content object is already available locally and may therefore not need to transmit a request for the object. As another example, high-priority content objects are removed from dynamic code features and statically injected into a content file (e.g., via a JavaScript). The end-user system then need not, e.g., process conditionalities pertaining to the content file. Thus, relevant portions of webpages may be quickly presented to users, and resource consumption may be reduced.

[0117] Referring next to FIG. 10, a block diagram of an embodiment of content-modifying system 1000 is shown. Any of a variety of components in the content distribution system 100 may include and/or be coupled to the content-modifying system 1000. For example, in some instances, a content-modifying system 1000 is implemented by one, more or each of a plurality of edge servers 230 (e.g., that respond to requests from users 128) at one, more or each of a plurality of POPs 120. Other implementations are possible. For example, the content-modifying system 1000 may be implemented by the origin server(s) 112, by one or more end user system(s) 102, by a server coupled to a plurality of edge servers 230 (e.g., at a POP), by a server coupled to a plurality of POPs 120, etc. In some instance, the content distribution system 100 includes a plurality of content-modifying systems, which may or may not be located in parallel locations (e.g., at a plurality of edge servers or at an edge server and at an origin server).

[0118] Content-modifying system 1000 includes a content- file receiver 1010. Content-file receiver 1010 receives a content file. In some instances, content- file receiver 1010 requests the content file (e.g., from an origin server) and receives the content file in response to the request. The content-file receiver 1010 may request the content file, e.g., upon detecting a request for the file and/or determining that the requested content file is not stored in a content files database 1020. A "content file", as used herein, may refer to a single file or a set of files. For example, a single HTML file may call or include additional files (e.g., style-sheet files and HTML files). A "content file" may thus, e.g., include all (e.g., markup and/or style-sheet) files included and cooperating to render content.

[0119] In some instances, content-file receiver 1010 receives the content file by

intercepting the content file while the content file is being transmitted to another location. Thus, for example, an edge server 230 may have requested the content file from an origin server, and a content-file receiver 1010 of a content-modifying system 1000 separate from the edge server 230 may detect a response to the request and intercept the content file.

[0120] In some instances, a content file is accessed from the content files database 1020. The content files database 1020 may include original and/or modified content files. Original and/or modified content files may be associated with storage locations, naming conventions, associated identifiers (e.g., in the content file or in an index of content files), etc. in order to convey a modification status. The content files database 1020 may or may not be stored locally with respect to the content- file modifier 1030. In some instances, one or more content files databases 1020 are located at each of a plurality of edge servers.

[0121] The content file may include a file written in a markup language (e.g., HTML) and/or a style-sheet language (e.g., Cascading Style Sheets). The content file may include one or more embedded or referenced content objects.

[0122] A content- file modifier 1030 modifies a content file (accessed via the content- file receiver 1010 or the content files database 1020). The content file may comprise a code, such as a code written in a markup language (e.g., HTML). The content file may be modified prior to a compiling of the content file. The content file may comprise a single file or a set of files. The content file may comprise a webpage-defining file.

[0123] The content-file modifier 1030 includes a reporting-code injector 1032, which injects a reporting code into the content file. The reporting code may include a reporting destination, reporting conditions and/or reporting protocols. For example, the reporting code may include instructions to send a report each time or a first time that a particular function is called, a type of function is called, specific/general content identified in the content file is retrieved from an external source, etc. Reporting protocols may include a requested frequency of reports. For example, reports may be sent every time one, more, a set or a threshold number of conditions are fulfilled; at fixed intervals but only when one, more, a set or a threshold number of conditions are fulfilled; at fixed intervals irrespective of condition fulfillment; etc. The reporting code may be injected, e.g., towards a beginning of the content file, near calls to functions, near embedded or referenced objects, etc.

[0124] The content-file modifier 1030 includes a content-file parser 1034, which parses the content file to identify one or more objects (e.g., embedded content objects, referenced content objects, code portions, included style sheets, etc.). For example, the content-file parser 1034 may identify embedded JavaScript scripts (e.g., by scanning for script tags), Flash animations, images, embedded links, etc. The objects may comprise, e.g., objects internal to the content file, external to the content file but associated with a same domain name as the content file (e.g., a link to a local image), or external to the content file and a hosting server.

[0125] The content-file parser 1034 may communicate with a stub controller 1036.

Operations of the content- file parser 1034 may be sequential (e.g., the content- file parser 1034 first identifying all objects and the stub controller 1036 then generating and/or injecting stubs) or not (e.g., the content- file parser 1034 identifying one or a subset of objects, the stub controller 1036 generating and/or injecting stubs for the identified objects, the content- file parser 1034 identifying another or another subset of objects, etc.).

[0126] After an content- file object is identified by the content-file parser 1034, the stub controller 1036 injects a stub into the content file or modifies an existing stub in the content file for the identified object. As used herein, a stub comprises a local piece of code that, when implemented, forwards a request for an object to a server and receive a response to the request. Thus, for example, a stub may request implementation of a remote JavaScript, receive the response, and present the response to a requesting user via, e.g., a webpage defined by the content file. As another example, a stub may request an image from a remote source, download the requested image, and present the image to a requesting user via, e.g., a webpage defined by the content file. A stub can be associated with (e.g., replace or represent), e.g., a whole script file, a part of a script file, one function, multiple functions, a part of a function, one object, or a part of an object.

[0127] Implementing the stub, therefore, comprises accessing a remote resource (e.g., using a remote JavaScript script, accessing a remote image file, etc.) and presenting either the remote resource or a result of the remote resource to a user. Thus, the user is unaware that the resource is remote and there is little to no functional content consequence of using a stub. However, because the accessed resource is remote, using a stub may delay presentations of content to a user.

[0128] The stub may be configured to marshal data (e.g., parameters passed to a remote server) and/or unmarshal data (e.g., in a received response to the request). The stub may comprise a JavaScript. The stub may be in communication with a skeleton which retrieves the requested object and transmit the requested object to the stub.

[0129] In some instances, one, more or all stubs comprise or are associated with a reporting feature. For example, the stub may include a piece of code which automatically transmits a report when the stub is called. As another example, the stub may be associated with a skeleton that transmits a report when the skeleton is called. In some instances, reporting features are not specifically integrated into a stub and/or skeleton, but the reporting code injected by the reporting-code injector 1032 may determine when specific stubs or stubs generally were called and report such calls.

[0130] In some instances, the injected reporting code or "null" stubs serve to identify when particular content objects are viewed or executed, but no request is sent for the content object. For example, a "null" stub or marker may be associated various HTML portions of a webpage code. When a user views or uses an associated webpage portion, a report may be generated and transmitted.

[0131] The stub controller 1036 comprises a stub generator 1038 configured to generate one or more stubs. In some instances, one stub is associated with each content object, or each content object of a particular type (e.g., a JavaScript, image, HTML portion, etc.) identified by the content-file parser. The stub generator 1038 may generate a stub, e.g., by generating another object (e.g., JavaScript). In some instances, the stub generator 1038 may further be configured to generate code to display placeholder content (e.g., stored locally) while the stub is collecting the requested content. For example, a placeholder small image file may be generated and/or associated with one or more content objects.

[0132] The stub controller 1036 further includes a stub injector 1040. The stub injector 1040 injects the generated code into the content file. Injecting the stub may include replacing a portion of the content file with the stub. For example, a function in the content file may be replaced with a stub. In some instances, injecting a stub comprises injecting and/or modifying code in one or more files other than the content file. For example, a skeleton code

corresponding to a stub may be generated to be stored in another file and/or at another server.

[0133] The modified content file may be stored, e.g., in a content-file database (such as content files database 1020) at an end user system, at a POP, at an edge server, at an origin server, etc. The modified content file may be stored in a local or remote location. For example, an edge server may comprise the content-modifying system 1000 and the modified content file may be stored locally. As another example, a server between an origin server and an edge server may comprise the content-modifying system 1000 and the modified content file may be stored at the edge server. In some instances, multiple versions of the modified content file are saved. For example, the modified content file may be generated at or near the origin server, and copies of the modified content file may then be stored at a plurality of (or each) POP within a CDN or at a plurality of (or each) edge server within a CDN. Copies may then be revised independently from each other and/or in a global manner. [0134] A content- file transmitter 1050 transmits the modified content file. The modified content file may be transmitted, e.g., to a POP, to an edge server, to an end user system, etc.

[0135] With reference to FIG. 11, a flow diagram of an embodiment of a process 1100 for generating a modified content file is shown. The depicted portion of the process begins in step 405 where the content-file modifier 1030 accesses one or more content files (e.g., received by the content- file receiver 1010 or accessed from a content files database 1020). A determination is made, at block 1110, as to whether the accessed content file is modified. For example, a determination may be made as to whether the content file comprises one or more usage-related reporting codes, whether one or more stubs are associated with the file, whether a modification identifier is associated with the file, etc. If the content file is modified, then then the process continues to block 1135, and the modified file is transmitted by the content- file transmitter 1050 (e.g., to an edge server, end user system, etc.).

[0136] If the content file is not modified, then processing continues to block 1115, at which reporting code is injected into the content file by the reporting-code injector 1032. The content- file parser 1034 parses the content file, at block 1120, to identify one or more content objects. Based on the identified content objects, a stub generator 1038 generates one or more stubs for each identified object at block 1125, and a stub injector 1040 injects the stubs into the content file at block 1130. Thus, the content file is modified to include one or more stubs that invoke remote resources and further include reporting functionality such that reports are generated and transmitted to indicate, e.g., which stubs are implemented and when they are implemented. The modified content file is then transmitted by the content-file transmitter 1050 (e.g., to an edge server, end user system, etc.) at block 1135.

[0137] With reference to FIG. 12, a flow diagram of an embodiment of a process 1200 for accessing content is shown. The depicted portion of the process begins in step 1205 where a content file (e.g., a webpage) is requested. For example, an user 128 may enter a uniform resource locator (URL), click on a hypertext link, etc. The request may be transmitted, e.g., from an end user system 102 associated with the user 128 over a network (e.g., the Internet 104) to a POP 120. The request may be assigned to an edge server 230 at the POP. [0138] In response to the request, a modified content file may be received by an end user system 102. For example, an edge server 230 may receive and modify the requested content file in accordance with process 1100 depicted in FIG. 11 and transmit the file to the end user system 102, or an origin server may access and modify the requested content file in accordance with process 1100 depicted in FIG. 11 and transmit the file to the end user system 102 via an edge server. In some instances, the end user system 102 receives an unmodified content file and itself modifies the file, e.g., in accordance with process 1100 depicted in FIG. 11.

[0139] At block 1215, the end user system 102 compiles at least part of the modified content file. At block 1220, a stub for a content object is invoked. For example, a user may attempt to interact with an applet or application, or a webpage may attempt to display an image. At block 1225, the stub invocation is reported. For example, the occurrence of the stub invocation, a time (e.g., relative to a request time, page-loading time, etc.) of the invocation, a result of the invocation (e.g., a response time), may be reported. In some instances, a report includes data about invocations of multiple stubs. Thus, generation and/or transmission of reports may be delayed, e.g., until a user has left a webpage.

[0140] At block 1230, the stub requests the content object or a result from the content object. For example, the stub may request that a JavaScript script compute a result based on user inputs, or the stub may request that the content object (e.g., an application or file) be delivered. The request may be sent, e.g., to a remote server, such as a server that provided the modified content file, an edge server, an origin server, etc. The request may be sent to a server that is within a same CDN as the requesting end user system 102 or outside CDN of the requesting end user system 102. [0141] In some instances, process 1200 does not include block 1220 and/or blocks 1230- 1235. For example, a reporting code in the modified content file may instruct the end user system 102 to report use of a content object (e.g., viewing a webpage portion associated with an HTML code portion or viewing a webpage portion affected by a style sheet), or null stubs may be inserted to report the use of a content object. In these embodiments, the used or viewed content object may already be available to end user system 102, such that it need not be requested.

[0142] In some instances, the request is sent to a skeleton (on a remote server, such as an edge server serving the content- file request)) associated with the invoked stub. The skeleton may then obtain the requested content object or result from the content object (e.g., from an origin server or CDN-external source). At block 1235, the content object or result from the content object is received from the stub. The content object or the result from the content object may be presented to the user 128. Blocks 1220-1235 may be repeated as other stubs are invoked.

[0143] Referring next to FIG. 13, a block diagram of an embodiment of components of the delivery controller 855 is shown. In this embodiment, the delivery controller 855 includes a bandwidth predictor 1305 configured to predict available bandwidth between a source location e.g., comprising the content-object database 850, and an end user system. The available bandwidth may be predicted by analyzing protocols, such as a Transmission Control Protocol, and limits associated with the protocol. The available bandwidth may be predicted based on, equations, transmissions of test objects, historical transmissions of non- test objects, etc.

[0144] A container generator 1310 generates one or more virtual containers, which may be stored in a container database 1320. A virtual container acts to group a plurality of content objects together. Thus, all of the group objects may be transmitted in a single transmission. The container generator 1310 may identify a number of virtual containers to generate based at least partly on the prediction of an available bandwidth made by the bandwidth predictor 1305. The number of containers may further depend on size of content objects and prioritizations of content objects. For example, if it is of high importance to fully transmit Content Objects #1 and #2 very quickly, but Content Object #3 is of less importance, it may be advantageous to generate two - not three - containers.

[0145] An object distributor 1315 distributes content objects amongst the generated containers. Distributing the containers may include determining which content objects are assigned to specific containers and further prioritizing and/or ordering content objects within a container. The distribution may depend upon prioritization of the content objects determined by the object prioritizor 345. In some embodiments, containers are configured such that multiple containers will be simultaneously transmitted (e.g., to an end user system 102). During the transmission of the container, content objects associated with a low rank (e.g., "1 ^st ") are transmitted prior to content objects associated with higher ranks. Thus, the object distributor 1315 may distribute the content objects such that high-priority content objects are distributed amongst the containers. This will allow for multiple high-priority content objects to be simultaneously transmitted over a network and to be quickly available to a user. [0146] In some instances, at least some containers include only content objects of a single type. For example, in one embodiment, at least one container only includes JavaScript content objects and/or only includes image objects. In these instances, separate containers may or may not include different types of content objects. For example, all containers may only include JavaScript content objects, or a first container may include only JavaScript content objects and a second container may include only image content objects.

[0147] In some instances, at least one container (or all containers) include or are configured to include content objects of a plurality of types. Thus, a single container may include, e.g., a Flash animation object, an image object, a style sheet, and a JavaScript. Embodiments disclosed herein further allow for prioritization of content objects across a plurality of types. Thus, it is possible to determine whether a particular JavaScript should receive a higher priority than a particular image object. This prioritization may be based at least partly on analyzing a frequency and/or timing of actual use of these objects.

[0148] In some instances, a container property (e.g., a number of containers, a distribution of content objects amongst the containers, which or how many content objects are distributed amongst the containers, etc.) is based on an analysis performed by a container-assignment analyzer 1330. The container-assignment analyzer 1330 analyzes a plurality of possible container assignment configurations. Each possible container assignment configuration may include a number of containers, and an assignment of each of a plurality of content objects to one of the containers. The container-assignment analyzer 1330 may estimate transmission properties associated with each assignment configuration.

[0149] The container-assignment analyzer 1330 includes a containerizing sample generator 1335 that generates a plurality of sample. Each sample includes a different container assignment configuration. Thus, one or more of the number of containers, container- assignment algorithms, number of assigned content objects and identity of assigned content objects may vary across the samples. For example, a first sample may include assigning Content Objects #1, #3 and #5 to Container #1 and Content Objects #2, #4 and #6 to

Container #2; a second sample may include assigning Content Objects #1, #3-6 to Container #1 and Content Object #2 and #7 to Container #2; and a third sample may include assigning Content Objects #1, #4 to Container #1, Content Object #2 to Container #2, and Content Object #3 and 5 to Container #3. In some instances, one or more of the number of containers, container-assignment algorithms, number of assigned content objects and identity of assigned content objects does not vary across the samples. For example, the container generator 1310 may determine a number of containers (e.g., independently from the container-assignment analyzer 1330 or based on a separate container-assignment analysis) and may transmit the number of containers to the container-assignment analyzer 1330. The number of containers may then remain fixed across generated samples. As another example, the number and/or identity of content objects to be assigned to containers may remain fixed across samples. [0150] A transmission-time estimator 1340 may estimate one or more transmission times for each sample. The estimated transmission time may include, e.g., an estimated

transmission time for one or more particular content objects (e.g., Content Object #1 will be transmitted in 2.5 seconds and Content Object #2 will be transmitted in 3.5 seconds, or Content Objects #1 and #2 will be transmitted in 3.5 seconds), for a group of content objects (e.g., all high priority content objects will be transmitted in 4.5 seconds), for all assigned content objects, etc. The transmission time may comprise a time relative to an initiation of container transmission, a request time, a response time, etc. The transmission times may be estimated, e.g., based on the bandwidth predicted by the bandwidth predictor 1305, empirical data analysis, size of content objects, etc. [0151] A sample scorer 1345 may determine a score for each sample. The score may include, e.g., a single value, multiple values, a vector, a distribution, a range, etc. The score may be numeric. The score may include a value along a finite or infinite continuum. The score may include a rank (e.g., ranking across samples). [0152] The score may be based at least partly on the estimated transmission time(s). For example, higher scores may be associated with shorter estimated transmission times. The score may depend on a total estimated transmission time (e.g., an estimated time for transmitting all containers or all content objects in containers), an estimated transmission time associated with a group of objects (e.g., an estimated time for transmitting all high- priority content objects), and/or an estimated transmission time associated with one, more or all particular objects. In one instance, a value for a transmission-time variable is determined for each content object of a plurality of content objects (e.g., all content objects to be transmitted), and the score is a weighted or unweighted sum of the values. [0153] In some instances, one or more use times are estimated by a use-time estimator 1350. A use time may identify a time at which it is estimated that a user will wish to use a content object (e.g., interact with a JavaScript script, view an image, initiate a Flash animation, etc.). The use time may be estimated, e.g., based on an analysis performed by the object usage analyzer 340, object usage statistics 825, data from the as-rendered locator 830, prioritizations determined by the object prioritizor 345, etc. For example, objects associated with a priority of sufficiently low rank may be associated with an estimated use time of "0", indicating that a user will likely wish to use the content object immediately upon loading a page. As another example, objects located at a top of a page may be associated with use times shorter than objects located at a bottom of a page. As another example, object use times may include a median actual use time. As another example, the object usage statistics may include a distribution of access times, and the use time may include a time at which X% (e.g., 5%) of the population had accessed the object by that point, or the use time may include a fastest access time amongst a population. The use-time estimation may include, e.g., a single time or distribution of times. [0154] The score determined by the sample scorer 1345 may at least partly depend on the one or more use times estimated by the use-time estimator 1350. In some instances, the score is at least partly based on comparing the estimated transmission time to the estimated use time. The comparison may result in a binary or non-binary value. For example, for each content object it may be determined whether the estimated transmission time is shorter than the estimated use time, or - in instances in which the estimated transmission time and/or estimated use time comprise probability distributions - the probability that the estimated transmission time is shorter than the estimated use time.

[0155] The container-assignment analyzer 1330 may include an object weight assigner 1355. The object weight assigner 1355 may assign a weight to one or more variables contributing to a score determined by the sample scorer 1345. In some instances, the score comprises a summation of variable values, each value being associated with a different content object or set of content objects. The object weight assigner 1355 may assign a weight to be associated with each content object or set of content objects. The weight may be based, e.g., on a priority determined by the object prioritizor 345. Thus, a score may be more heavily influenced by estimated transmission times associated with high-priority content objects as compared to estimated transmission times associated with low-priority content objects.

[0156] Based on the scores determined by the sample scorer, the container generator 1310 may determine a number of containers to generator and/or the object distributor 1315 may determine a distribution scheme, a number of content objects to distribute amongst containers and/or an identity of content objects to distribute amongst containers. The determination may be based at least partly on samples associated with high scores. For example, if a highest score sample has a particular content-object distribution technique, the object distributor 1315 may use a similar distribution technique. [0157] In some instances, the delivery controller 855 does not include a container- assignment analyzer 1330 that generates and analyzes samples. For example, the delivery controller 855 may determine a number of containers to be generated and/or object distribution schemes based on solving one or more transmission equations to transmit select or all of the content objects quickly, most quickly or prior to estimated use. [0158] With reference to FIG. 14, a flow diagram of an embodiment of a process 1400 for delivering content objects is shown. The depicted portion of the process begins in step 1405 where a transmission protocol is analyzed. For example, a protocol, such as TCP, may limit throughput of a connection based on various factors. These factors may be identified and analyzed. [0159] At block 1410, the bandwidth predictor 1305 predicts the a bandwidth. At block

1415, a container generator determines a number of containers to associate with a content file or a number of containers generally. The number of containers may be based at least partly on the predicted bandwidth and analysis of the transmission protocol. At block 1420, the container generator generates the containers. At block 1425, the object distributor 1315 receives object prioritization e.g., from the object prioritizor 345. At block 1430, an object- distribution scheme is determined by the object distributor 1315. The object-distribution scheme may or may not vary across different distributions. A distribution scheme may consider, e.g., object priorities, a number of generated containers, object sizes, and/or estimated object use times to determine how objects are to be distributed amongst the containers. At block 1435, content objects (e.g., associated with a content file) are distributed amongst the containers in accordance with the determined object-distribution scheme. At block 1440, the containers are then be delivered, e.g., to an end user system 102. [0160] Thus, each content file of a plurality of content files may be associated with one or more (e.g., a plurality of) containers, and at least some of the content objects associated with the content file may be distributed amongst the containers. In some instances, at least some content objects are not distributed amongst the containers. For example, some (e.g., high priority) content objects may be integrated into the content file itself, and/or some (e.g., rarely used) content objects may be available only upon request.

[0161] With reference to FIG. 15, a flow diagram of an embodiment of a process 1500 for determining a container property (e.g., a number of containers to generate, an object- distribution scheme, etc.) is shown. The depicted portion of the process begins in step 1505 where a containerizing sample is generated by the containerizing sample generator 1335. The sample may include a container assignment for each of a plurality of content objects. At block 1510, one or more transmission times are estimated by the transmission-time estimator 1340. For example, an estimate of a time required for each content object of the plurality of content objects to be fully transmitted, relative to an initiation of container transmission, may be made. At block 1515, one or more use times are estimated by use-time estimator 1350. The use-time estimates may depend, e.g., on an empirical-data analysis, as-rendered page location of the objects, etc. At block 1520, a weight is determined for one or more content objects or groups of content objects by the object weight assigner 1355. For example, a weight may be determined for each content object to be transmitted. The weight may depend, e.g., on a priority associated with the content object. At block 1525, a score is assigned to the sample by the sample scorer 1345. The assigned score may be based on one or more of the estimated transmission time(s), estimated use time(s), and content-object weight(s). For example, in one instance, the assigned score depends on the number of content objects estimated to be fully transmitted prior to a user's attempted use of the object.

[0162] Blocks 1505-1525 may be repeated, such that a plurality of samples are generated and scored. Each sample may be associated, e.g., with different distributions of content objects and/or number of containers. At block 1530, a container property (e.g., number of containers, distribution schemes, content objects to distribute, etc.) is determined. The determined container property may or may not be specific, e.g., to particular content objects, to a specific content file embedding or referencing the content objects, specific transmission lines, specific requests, etc.

[0163] Referring next to FIG. 16, an exemplary environment with which embodiments may be implemented is shown with a computer system 1600 that can be used by a designer 1604 to design, for example, electronic designs. The computer system 1600 can include a computer 1602, keyboard 1622, a network router 1612, a printer 1608, and a monitor 1606. The monitor 1606, processor 1602 and keyboard 1622 are part of a computer system 1626, which can be a laptop computer, desktop computer, handheld computer, mainframe computer, etc. The monitor 1606 can be a CRT, flat screen, etc. [0164] A designer 1604 can input commands into the computer 1602 using various input devices, such as a mouse, keyboard 1622, track ball, touch screen, etc. If the computer system 1600 comprises a mainframe, a designer 1604 can access the computer 1602 using, for example, a terminal or terminal interface. Additionally, the computer system 1626 may be connected to a printer 1608 and a server 1610 using a network router 1612, which may connect to the Internet 1618 or a WAN.

[0165] The server 1610 may, for example, be used to store additional software programs and data. In one embodiment, software implementing the systems and methods described herein can be stored on a storage medium in the server 1610. Thus, the software can be run from the storage medium in the server 1610. In another embodiment, software implementing the systems and methods described herein can be stored on a storage medium in the computer 1602. Thus, the software can be run from the storage medium in the computer system 1626. Therefore, in this embodiment, the software can be used whether or not computer 1602 is connected to network router 1612. Printer 1608 may be connected directly to computer 1602, in which case, the computer system 1626 can print whether or not it is connected to network router 1612.

[0166] With reference to FIG. 17, an embodiment of a special-purpose computer system 1700 is shown. For example, one or more of content-modifying system 1000, acceleration engine 330, delivery controller 855 may be a special-purpose computer system 1700. The above methods may be implemented by computer-program products that direct a computer system to perform the actions of the above-described methods and components. Each such computer-program product may comprise sets of instructions (codes) embodied on a computer-readable medium that directs the processor of a computer system to perform corresponding actions. The instructions may be configured to run in sequential order, or in parallel (such as under different processing threads), or in a combination thereof. After loading the computer-program products on a general purpose computer system 1626, it is transformed into the special -purpose computer system 1700. [0167] Special-purpose computer system 1700 comprises a computer 1602, a monitor 1606 coupled to computer 1602, one or more additional user output devices 1730

(optional) coupled to computer 1602, one or more user input devices 1740 (e.g., keyboard, mouse, track ball, touch screen) coupled to computer 1602, an optional communications interface 1750 coupled to computer 1602, a computer-program product 1705 stored in a tangible computer-readable memory in computer 1602. Computer- program product 1705 directs system 1700 to perform the above-described methods. Computer 1602 may include one or more processors 1760 that communicate with a number of peripheral devices via a bus subsystem 1790. These peripheral devices may include user output device(s) 1730, user input device(s) 1740, communications interface 1750, and a storage subsystem, such as random access memory (RAM) 1770 and nonvolatile storage drive 1780 (e.g., disk drive, optical drive, solid state drive), which are forms of tangible computer-readable memory.

[0168] Computer-program product 1705 may be stored in non- volatile storage drive 1780 or another computer-readable medium accessible to computer 1602 and loaded into memory 1770. Each processor 1760 may comprise a microprocessor, such as a microprocessor from Intel ^® or Advanced Micro Devices, Inc. ^® , or the like. To support computer-program product 1705, the computer 1602 runs an operating system that handles the communications of product 1705 with the above-noted components, as well as the communications between the above-noted components in support of the computer- program product 1705. Exemplary operating systems include Windows ^® or the like from Microsoft Corporation, Solaris ^® from Sun Microsystems, LINUX, UNIX, and the like.

[0169] User input devices 1740 include all possible types of devices and mechanisms to input information to computer system 1626. These may include a keyboard, a keypad, a mouse, a scanner, a digital drawing pad, a touch screen incorporated into the display, audio input devices such as voice recognition systems, microphones, and other types of input devices. In various embodiments, user input devices 1740 are typically embodied as a computer mouse, a trackball, a track pad, a joystick, wireless remote, a drawing tablet, a voice command system. User input devices 1740 typically allow a user to select objects, icons, text and the like that appear on the monitor 1606 via a command such as a click of a button or the like. User output devices 1730 include all possible types of devices and mechanisms to output information from computer 1602. These may include a display (e.g., monitor 1606), printers, non- visual displays such as audio output devices, etc.

[0170] Communications interface 1750 provides an interface to other communication networks and devices and may serve as an interface to receive data from and transmit data to other systems, WANs and/or the Internet 1618. Embodiments of communications interface 1750 typically include an Ethernet card, a modem (telephone, satellite, cable, ISDN), a (asynchronous) digital subscriber line (DSL) unit, a FireWire ^® interface, a

USB ^® interface, a wireless network adapter, and the like. For example, communications interface 1750 may be coupled to a computer network, to a FireWire ^® bus, or the like. In other embodiments, communications interface 1750 may be physically integrated on the motherboard of computer 1602, and/or may be a software program, or the like. [0171] RAM 1770 and non- volatile storage drive 1780 are examples of tangible computer-readable media configured to store data such as computer-program product embodiments of the present invention, including executable computer code, human- readable code, or the like. Other types of tangible computer-readable media include floppy disks, removable hard disks, optical storage media such as CD-ROMs, DVDs, bar codes, semiconductor memories such as flash memories, read-only-memories (ROMs), battery-backed volatile memories, networked storage devices, and the like. RAM 1770 and non-volatile storage drive 1780 may be configured to store the basic programming and data constructs that provide the functionality of various embodiments of the present invention, as described above. [0172] Software instruction sets that provide the functionality of the present invention may be stored in RAM 1770 and non- volatile storage drive 1780. These instruction sets or code may be executed by the processor(s) 1760. RAM 1770 and non- volatile storage drive 1780 may also provide a repository to store data and data structures used in accordance with the present invention. RAM 1770 and non-volatile storage drive 1780 may include a number of memories including a main random access memory (RAM) to store of instructions and data during program execution and a read-only memory (ROM) in which fixed instructions are stored. RAM 1770 and non- volatile storage drive 1780 may include a file storage subsystem providing persistent (non-volatile) storage of program and/or data files. RAM 1770 and non- volatile storage drive 1780 may also include removable storage systems, such as removable flash memory.

[0173] Bus subsystem 1790 provides a mechanism to allow the various components and subsystems of computer 1602 communicate with each other as intended. Although bus subsystem 1790 is shown schematically as a single bus, alternative embodiments of the bus subsystem may utilize multiple busses or communication paths within the computer 1602.

[0174] Specific details are given in the above description to provide a thorough understanding of the embodiments. However, it is understood that the embodiments may be practiced without these specific details. For example, circuits may be shown in block diagrams in order not to obscure the embodiments in unnecessary detail. In other instances, well-known circuits, processes, algorithms, structures, and techniques may be shown without unnecessary detail in order to avoid obscuring the embodiments. [0175] Implementation of the techniques, blocks, steps and means described above may be done in various ways. For example, these techniques, blocks, steps and means may be implemented in hardware, software, or a combination thereof. For a hardware

implementation, the processing units may be implemented within one or more application specific integrated circuits (ASICs), digital signal processors (DSPs), digital signal processing devices (DSPDs), programmable logic devices (PLDs), field programmable gate arrays (FPGAs), processors, controllers, micro-controllers, microprocessors, other electronic units designed to perform the functions described above, and/or a combination thereof.

[0176] Also, it is noted that the embodiments may be described as a process which is depicted as a flowchart, a flow diagram, a data flow diagram, a structure diagram, or a block diagram. Although a flowchart may describe the operations as a sequential process, many of the operations can be performed in parallel or concurrently. In addition, the order of the operations may be re-arranged. A process is terminated when its operations are completed, but could have additional steps not included in the figure. A process may correspond to a method, a function, a procedure, a subroutine, a subprogram, etc. When a process corresponds to a function, its termination corresponds to a return of the function to the calling function or the main function. [0177] Furthermore, embodiments may be implemented by hardware, software, scripting languages, firmware, middleware, microcode, hardware description languages, and/or any combination thereof. When implemented in software, firmware, middleware, scripting language, and/or microcode, the program code or code segments to perform the necessary tasks may be stored in a machine readable medium such as a storage medium. A code segment or machine-executable instruction may represent a procedure, a function, a subprogram, a program, a routine, a subroutine, a module, a software package, a script, a class, or any combination of instructions, data structures, and/or program statements. A code segment may be coupled to another code segment or a hardware circuit by passing and/or receiving information, data, arguments, parameters, and/or memory contents.

Information, arguments, parameters, data, etc. may be passed, forwarded, or transmitted via any suitable means including memory sharing, message passing, token passing, network transmission, etc.

[0178] For a firmware and/or software implementation, the methodologies may be implemented with modules (e.g., procedures, functions, and so on) that perform the functions described herein. Any machine-readable medium tangibly embodying instructions may be used in implementing the methodologies described herein. For example, software codes may be stored in a memory. Memory may be implemented within the processor or external to the processor. As used herein the term "memory" refers to any type of long term, short term, volatile, nonvolatile, or other storage medium and is not to be limited to any particular type of memory or number of memories, or type of media upon which memory is stored.

[0179] Moreover, as disclosed herein, the term "storage medium" may represent one or more memories for storing data, including read only memory (ROM), random access memory (RAM), magnetic RAM, core memory, magnetic disk storage mediums, optical storage mediums, flash memory devices and/or other machine readable mediums for storing information. The term "machine-readable medium" includes, but is not limited to portable or fixed storage devices, optical storage devices, wireless channels, and/or various other storage mediums capable of storing that contain or carry instruction(s) and/or data.

[0180] While the principles of the disclosure have been described above in connection with specific apparatuses and methods, it is to be clearly understood that this description is made only by way of example and not as limitation on the scope of the disclosure.