IN A DATA CENTER FABRIC

Technical Field

[0001] The disclosure provides a method of creating a virtual netflow collector in which netflow packet collection is distributed across switches in an Application Centric Infrastructure fabric, a hash combination is calculated for packet subflows and the hash combination is mapped to an IP multicast address for mapping to a physical netflow collector.

Background

[0002] Today, netflow data collection is on a per node, per interface basis and is configured and managed for individual switches. The current approach has the following limitations. First, it is very difficult to correlate common (such as Tenant, Context (virtual routing and forwarding or VRF), Bridge Domains (BD)) and granular statistics (Application stats) across a network of switches, unless all the flow statistics go to the same collector. In a typical Application Centric Infrastructure (ACI) deployment, collecting fabric-level netflow statistics on virtual constructs such as the Tenant, VRF or BD is difficult, as the flows for these higher- level constructs will be spread across multiple switches in the fabric and these switches may be using different collectors for bandwidth scaling. Also, in a controller managed datacenter fabric, it is desired to collect finer statistics at various scopes than a traditional network. For instance, an administrator might want to collect statistics of a particular application for a tenant and multiple instances of this application can be running attached to different switches in the fabric. In general, fabric-wide granular netflow support will help provide meaningful information of application flows in the world of ACI. [0003] Another limitation is the scalability and the management of the netflow collectors which cater to these set of the switches. In the current method, a flow collector is statically mapped to a netflow monitoring entity such as an interface on a switch. This method cannot scale when the bandwidth needs are different across different interfaces or switches. Also, when there are more collectors/switches, it becomes too difficult to manage the collector configuration. In a Dynamic Virtual Machine (VM) management environment, a collector should be able to cater to the VM moves. The same collector has to be provisioned across the entire domain where the VM could move.

[0004] As ACI ventures into cloud deployments, the requirement for an efficient netflow solution is even more compelling, as the fabric will be extended to support higher scale of virtual leaf switches and virtual PODs in the cloud. In this environment, managing netflow collectors per virtual leaf and maintaining a large number of collectors will be difficult.

BRIEF DESCRIPTION OF THE DRAWINGS

[0005] In order to describe the manner in which the above-recited and other advantages and features of the disclosure can be obtained, a more particular description of the principles briefly described above will be rendered by reference to specific embodiments thereof which are illustrated in the appended drawings.

Understanding that these drawings depict only exemplary embodiments of the disclosure and are not therefore to be considered to be limiting of its scope, the principles herein are described and explained with additional specificity and detail through the use of the accompanying drawings in which:

[0006] FIG. 1 illustrates an example system configuration;

[0007] FIG. 2 illustrates an example network environment; [0008] FIG. 3 illustrates a collector cluster and numerous respective collectors connected thereto;

[0009] FIG. 4 illustrates further details associated with the collector cluster and the approach disclosed herein;

[0010] FIG. 5 illustrates a method embodiment; and

[0011] FIG. 6 illustrates another method embodiment.

DESCRIPTION OF EXAMPLE EMBODIMENTS

[0012] Various embodiments of the disclosure are discussed in detail below. While specific implementations are discussed, it should be understood that this is done for illustration purposes only. A person skilled in the relevant art will recognize that other components and configurations may be used without parting from the spirit and scope of the disclosure.

Overview

[0013] Additional features and advantages of the disclosure will be set forth in the description which follows, and in part will be obvious from the description, or can be learned by practice of the herein disclosed principles. The features and advantages of the disclosure can be realized and obtained by means of the instruments and combinations particularly pointed out in the appended claims. These and other features of the disclosure will become more fully apparent from the following description and appended claims, or can be learned by the practice of the principles set forth herein.

[0014] A method aspect of this disclosure includes distributing netflow packet collectors across all switched in a network fabric of switches to yield a virtual netflow collector, calculating a first hash at a source switch of an incoming packet on each possible combination of a 5-tuple IP packet header, calculating a second hash at the source switch of an incoming packet on each possible combination of at least one virtual attribute of the network fabric of switches, creating an N-Tuple flow from the first hash and the second hash, exporting the N-Tuple flow to the virtual netflow collector and mapping, via a virtual extensible local area network multicast address group, the virtual netflow collector to one or more physical netflow collector.

[0015] Another method aspect includes calculating, at a collector receiving a data flow and via a hashing algorithm, all possible hashes associated with at least one virtual attribute associated with the data flow to yield resultant hash values, and, based on the resultant hash values, computing a multicast address group and multicasting the data flow to n leafs based on the multicast address group. The method includes, at respective other collectors, filtering received sub-flows of the data flow based on the resultant hashes, wherein if a respective collector is owned by a hash, the respective collector accepts and saves the sub-flow in a local switch collector database. The method also can include receiving a query using the hashing algorithm to query a relevant aggregated or granular flow.

Detailed Description

[0016] The present disclosure addresses the issues raised above. The disclosure provides a system, method and computer-readable storage device embodiments. The concepts disclosed herein address the monitoring requirements for a high scale datacenter fabric environment. The concepts include correlating common attributes across a network of switches, providing a granular view of statistics which makes any form of visualization and projection easy, providing a scalable collection mechanism which can elastically handle additional nodes and bandwidth, and placing a collector that is decoupled from the monitoring entity, which helps dynamically migrating the collector to any node without having to change anything in the monitoring entity. These features allow the netflow collection to be placed through any workload orchestration, making better use of distributed compute resources.

[0017] First a general example system shall be disclosed in FIG. 1 which can provide some basic hardware components making up a server, node or other computer system. FIG. 1 illustrates a computing system architecture 100 wherein the components of the system are in electrical communication with each other using a connector 105.

Exemplary system 100 includes a processing unit (CPU or processor) 110 and a system connector 105 that couples various system components including the system memory 115, such as read only memory (ROM) 120 and random access memory (RAM) 125, to the processor 110. The system 100 can include a cache of high-speed memory connected directly with, in close proximity to, or integrated as part of the processor 110. The system 100 can copy data from the memory 115 and/or the storage device 130 to the cache 112 for quick access by the processor 110. In this way, the cache can provide a performance boost that avoids processor 110 delays while waiting for data. These and other modules/services can control or be configured to control the processor 110 to perform various actions. Other system memory 115 may be available for use as well. The memory 115 can include multiple different types of memory with different performance characteristics. The processor 110 can include any general purpose processor and a hardware module or software module/service, such as service 1 132, service 2 134, and service 3 136 stored in storage device 130, configured to control the processor 110 as well as a special- purpose processor where software instructions are incorporated into the actual processor design. The processor 110 may essentially be a completely self-contained computing system, containing multiple cores or processors, a bus (connector), memory controller, cache, etc. A multi-core processor may be symmetric or asymmetric.

[0018] To enable user interaction with the computing device 100, an input device 145 can represent any number of input mechanisms, such as a microphone for speech, a touch-sensitive screen for gesture or graphical input, keyboard, mouse, motion input, speech and so forth. An output device 135 can also be one or more of a number of output mechanisms known to those of skill in the art. In some instances, multimodal systems can enable a user to provide multiple types of input to communicate with the computing device 100. The communications interface 140 can generally govern and manage the user input and system output. There is no restriction on operating on any particular hardware arrangement and therefore the basic features here may easily be substituted for improved hardware or firmware arrangements as they are developed.

[0019] Storage device 130 is a no n- volatile memory and can be a hard disk or other types of computer readable media which can store data that are accessible by a computer, such as magnetic cassettes, flash memory cards, solid state memory devices, digital versatile disks, cartridges, random access memories (RAMs) 125, read only memory (ROM) 120, and hybrids thereof.

[0020] The storage device 130 can include software services 132, 134, 136 for controlling the processor 110. Other hardware or software modules/services are contemplated. The storage device 130 can be connected to the system connector 105. In one aspect, a hardware module that performs a particular function can include the software component stored in a computer-readable medium in connection with the necessary hardware components, such as the processor 110, connector 105, display 135, and so forth, to carry out the function. [0021] Having introduced the basic computing components which can be applicable to embodiments associated with this disclosure, the disclosure now turn to FIG. 2 which illustrates an example network environment.

[0022] FIG. 2 illustrates a diagram of example network environment 200. This figure is discussed with the concept of capturing agents on various network components. It is noted that the disclosed concept discussed below with respect to FIG. 4, and the focus of this disclosure, differs in how collection is done from what is referenced in FIG. 2. With reference to FIG. 2, fabric 212 can represent the underlay (i.e., physical network) of network environment 200. Fabric 212 can include spine routers 1-N(202 _A - _N ) (collectively "202") and leaf routers 1-N(204 _A - _N ) (collectively "204"). Leaf routers 204 can reside at the edge of fabric 212, and can thus represent the physical network edges. Leaf routers 204 can be, for example, top-of-rack ("ToR") switches, aggregation switches, gateways, ingress and/or egress switches, provider edge devices, and/or any other type of routing or switching device.

[0023] Leaf routers 204 can be responsible for routing and/or bridging tenant or endpoint packets and applying network policies. Spine routers 202 can perform switching and routing within fabric 212. Thus, network connectivity in fabric 212 can flow from spine routers 202 to leaf routers 204, and vice versa.

[0024] Leaf routers 204 can provide servers 1-5 (206 _A -E) (collectively "206"), hypervisors 1-4 (208 _A -208 _D ) (collectively "208"), and virtual machines (VMs) 1-5 (210 _A -210 _E ) (collectively "210") access to fabric 212. For example, leaf routers 204 can encapsulate and decapsulate packets to and from servers 206 in order to enable communications throughout environment 200. Leaf routers 204 can also connect other devices, such as device 214, with fabric 212. Device 214 can be any network- capable device(s) or network(s), such as a firewall, a database, a server, a collector 218 (further described below), an engine 220 (further described below), etc. Leaf routers 204 can also provide any other servers, resources, endpoints, external networks, VMs, services, tenants, or workloads with access to fabric 212.

[0025] VMs 210 can be virtual machines hosted by hypervisors 208 running on servers 206. VMs 210 can include workloads running on a guest operating system on a respective server. Hypervisors 208 can provide a layer of software, firmware, and/or hardware that creates and runs the VMs 210. Hypervisors 208 can allow VMs 210 to share hardware resources on servers 206, and the hardware resources on servers 206 to appear as multiple, separate hardware platforms. Moreover, hypervisors 208 and servers 206 can host one or more VMs 210. For example, server 206 _A and hypervisor 208 _A can host VMs 210 _A _ _B .

[0026] In some cases, VMs 210 and/or hypervisors 208 can be migrated to other servers 206. For example, VM 210 _A can be migrated to server 206c and hypervisor 208B. Servers 206 can similarly be migrated to other locations in network environment 200. A server connected to a specific leaf router can be changed to connect to a different or additional leaf router. In some cases, some or all of servers 206, hypervisors 208, and/or VMs 210 can represent tenant space. Tenant space can include workloads, services, applications, devices, and/or resources that are associated with one or more clients or subscribers. Accordingly, traffic in network environment 200 can be routed based on specific tenant policies, spaces, agreements,

configurations, etc. Moreover, addressing can vary between one or more tenants. In some configurations, tenant spaces can be divided into logical segments and/or networks and separated from logical segments and/or networks associated with other tenants. [0027] Any of leaf routers 204, servers 206, hypervisors 208, and VMs 210 can include a capturing agent (also referred to as a "sensor") configured to capture network data, and report any portion of the captured data to collector 218. Capturing agents 216 can be processes, agents, modules, drivers, or components deployed on a respective system (e.g., a server, VM, hypervisor, leaf router, etc.), configured to capture network data for the respective system (e.g., data received or transmitted by the respective system), and report some or all of the captured data to collector 218.

[0028] For example, a VM capturing agent can run as a process, kernel module, or kernel driver on the guest operating system installed in a VM and configured to capture data (e.g., network and/or system data) processed (e.g., sent, received, generated, etc.) by the VM. Additionally, a hypervisor capturing agent can run as a process, kernel module, or kernel driver on the host operating system installed at the hypervisor layer and configured to capture data (e.g., network and/or system data) processed (e.g., sent, received, generated, etc.) by the hypervisor. A server capturing agent can run as a process, kernel module, or kernel driver on the host operating system of a server and configured to capture data (e.g., network and/or system data) processed (e.g., sent, received, generated, etc.) by the server. And a network device capturing agent can run as a process or component in a network device, such as leaf routers 204, and configured to capture data (e.g., network and/or system data) processed (e.g., sent, received, generated, etc.) by the network device.

[0029] Capturing agents or sensors can be configured to report the observed data and/or metadata about one or more packets, flows, communications, processes, events, and/or activities to collector 218. For example, capturing agents can capture network data as well as information about the system or host of the capturing agents (e.g., where the capturing agents are deployed). Such information can also include, for example, data or metadata of active or previously active processes of the system, operating system user identifiers, metadata of files on the system, system alerts, networking information, etc. Capturing agents may also analyze all the processes running on the respective VMs, hypervisors, servers, or network devices to determine specifically which process is responsible for a particular flow of network traffic. Similarly, capturing agents may determine which operating system user(s) is responsible for a given flow. Reported data from capturing agents can provide details or statistics particular to one or more tenants. For example, reported data from a subset of capturing agents deployed throughout devices or elements in a tenant space can provide information about the performance, use, quality, events, processes, security status, characteristics, statistics, patterns, conditions, configurations, topology, and/or any other information for the particular tenant space.

[0030] Collectors 218 can be one or more devices, modules, workloads and/or processes capable of receiving data from capturing agents. Collectors 218 can thus collect reports and data from capturing agents. Collectors 218 can be deployed anywhere in network environment 200 and/or even on remote networks capable of communicating with network environment 200. For example, one or more collectors can be deployed within fabric 212 or on one or more of the servers 206. One or more collectors can be deployed outside of fabric 212 but connected to one or more leaf routers 204. Collectors 218 can be part of servers 206 and/or separate servers or devices (e.g., device 214). Collectors 218 can also be implemented in a cluster of servers.

[0031] Collectors 218 can be configured to collect data from capturing agents. In addition, collectors 218 can be implemented in one or more servers in a distributed fashion. As previously noted, collectors 218 can include one or more collectors. Moreover, each collector can be configured to receive reported data from all capturing agents or a subset of capturing agents. For example, a collector can be assigned to a subset of capturing agents so the data received by that specific collector is limited to data from the subset of capturing agents.

[0032] Collectors 218 can be configured to aggregate data from all capturing agents and/or a subset of capturing agents. Moreover, collectors 218 can be configured to analyze some or all of the data reported by capturing agents. For example, collectors 218 can include analytics engines (e.g., engines 220) for analyzing collected data. Environment 200 can also include separate analytics engines 220 configured to analyze the data reported to collectors 218. For example, engines 220 can be configured to receive collected data from collectors 218 and aggregate the data, analyze the data (individually and/or aggregated), generate reports, identify conditions, compute statistics, visualize reported data, present visualized data, troubleshoot conditions, visualize the network and/or portions of the network (e.g., a tenant space), generate alerts, identify patterns, calculate misconfigurations, identify errors, generate suggestions, generate testing, and/or perform any other analytics functions.

[0033] While collectors 218 and engines 220 are shown as separate entities, this is for illustration purposes as other configurations are also contemplated herein. For example, any of collectors 218 and engines 220 can be part of a same or separate entity. Moreover, any of the collector, aggregation, and analytics functions can be implemented by one entity (e.g., collectors 218) or separately implemented by multiple entities (e.g., engine 220 and/or collectors 218).

[0034] Each of the capturing agents can use a respective address (e.g., internet protocol (IP) address, port number, etc.) of their host to send information to collectors 218 and/or any other destination. Collectors 218 may also be associated with their respective addresses such as IP addresses. Moreover, capturing agents can periodically send information about flows they observe to collectors 218. Capturing agents can be configured to report each and every flow they observe. Capturing agents can report a list of flows that were active during a period of time (e.g., between the current time and the time of the last report). The consecutive periods of time of observance can be represented as pre-defined or adjustable time series. The series can be adjusted to a specific level of granularity. Thus, the time periods can be adjusted to control the level of details in statistics and can be customized based on specific requirements, such as security, scalability, storage, etc. The time series information can also be implemented to focus on more important flows or components (e.g., VMs) by varying the time intervals. The communication channel between a capturing agent and collector 218 can also create a flow in every reporting interval. Thus, the information transmitted or reported by capturing agents can also include information about the flow created by the communication channel.

[0035] FIG. 3 illustrates a network fabric having a Leaf 1 (302) receiving a flow fi , Leaf 2 (304) receiving a flow f2, Leaf 3 (306) receiving a flow f ₃ and a Leaf n (308) receiving a flow f _n . The flows fi , f2,f3 and f _n each represent the same flow (5-tuple) (or different flows) coming into different switches. For example, the flow could be into the same bridge domain (BD) deployed on n switches.

[0036] The features fi. _te nant, f2.tenant, and f3.tenant represent an example of the tenant level sub-flow created for a given tenant, such as fi. tenants- 1 · A tenant is a logical container for application policies that enable an administrator to exercise domain-based access control. The features fi.tenant.vrf, f2.tenant.wf, and f3.tenant.wf each represent an example of the tenant+vrf- level sub-flow created for a given virtual routing and forwarding (VRF) object (or context) which is a tenant network. A VRF is a unique layer 3 forwarding an application policy domain. For example, the VRF can be

characterized as tn- 1 :vrf- 1.

[0037] The features fi .tenant, wfhttps, f ₂ .tenant.vrf.http _S , and f3.tenant.vrf.http _S each represent the https sub-flow for tn-l :vrf-l . As shown in FIG. 3, each of the 5-tuple data flows into a respective leaf node. A 5-tuple refers to a set of five different values that are a part of a Transmission Control Protoeol htternet Protocol (TCP/IP) connection. It includes a source IP address/port number, destination IP address/port number and the protocol in use. From the packet headers and incoming interface on the switch (Leaf 1 , Leaf 2, etc.), each respective switch can derive other virtual attributes like Tenant, VRF, BD, Application, Endpoint Group (EPG) and create finer flows (N Tuples) from the initial 5-tuple flow. Examples of the sub-flows or finer flows are shown as the fi.tenant, fi.tenant.vrf and fi.tenant.vrf.https flows from Leaf I. Other suh-fiows can of course be derived as well. Each of these sub/micro flows corresponds to a combination of one or more of the attributes of the 5-tuple along with one or more of the virtual attributes.

[0038] Each of the sub-flows is exported to the virtual netflow collector or collector cluster 310, which gets mapped in the network to one or more of the physical netflow collectors 312, 314, 316 in the cluster through consistent hashing of the sub-flow parameters. As is shown in FIG. 3, all of the f _x . _te nant sub-flows are mapped to collector 1 (312), all of the f _x . _te nant.vrf sub-flows are mapped to collector 2 (314) and all of the fi.tenant. wf.https sub-flows are mapped to collector n (316).

[0039] A given sub-flow created in one or more switches (302, 304, 306, 308) in the fabric will end up in the same physical instance of the collector cluster to provide a aggregated view. As an example, HTTPS traffic statistics on a given BD which is spread across multiple switches will always end up in the same physical collector instance. Similarly, aggregated traffic stats for a VRF will end in one collector instance.

[0040] The same idea applies for any visualization of the collected flows. Since the sub-flow is mapped to a physical collector inside the virtual collector 310, any form of query can be targeted at the virtual collector 310, which then gets internally mapped to the physical collector instance (312, 314, 316) holding the sub-flow. For example, an administrator can query the system for "https traffic for BD bl in tenant tl " and can automatically be redirected by the network to the collector instance holding the entry, which is collector n (316) in FIG. 3. In one aspect, the network fabric presents one collector view for both collection and visualization.

[0041] While the above steps can be applied for any network of switches that are logically managed together, it is particularly significant in the ACI fabric. In the ACI fabric, the netflow collector functionality can be distributed across the spine and leaf switches and the underlay network control and data plane can be leveraged to provide the network function that maps the sub-flow to the physical collector instance. The entire ACI fabric can be envisioned as one netflow monitor and collector domain, which can also provide visualization interface through the REST interface. The REST interface is the Representative State Transfer and it relies on a stateless, client- server, cacheable communications protocol.

[0042] The solution disclosed herein makes use of the fabric compute resources like the leaf and spine CPU resources to run collector functionality, instead of mandating external collector nodes. Also, the solution uses the underlay multicast and VxLAN (discussed more fully below) segmentation to map and efficiently deliver the sub- flows to the corresponding collectors residing in the Leaf/Spine switches. This mapping function is done in a deterministic and distributed way without need for synchronization. This helps the leaf and spine switches that form the collector domain to be dynamically rearranged based on the available compute resources to do the functionality. When a leaf leaves the collector domain (or) when a new leaf gets added to the collector domain, the collection functionality in the rest of the fabric is not affected. This provides one of the benefits of the present disclosure, which is the ability to scale or expand easily for new leaf nodes added to a fabric.

[0043] Virtual Extensible local area network (VxLAN) is a network

virtu alization technology that attempts to address the scalability problems associated with large cloud computing deployments. The technology applies a virtual LAN (VLAN) type of encapsulation technique to encapsulate MAC-based OSI layer 2 Ethernet frames within layer 4 UDP (user datagram protocol) packets, using 4789 as the default IANA-assigned destination UDP port number. VxLAN endpoints, which terminate VxLAN tunnels and may be virtual and/or physical switch ports, are known as VxLAN tunnel endpoints (VTEPs).

[0044] VxLAN seeks to standardize on an overlay encapsulation

protocol. Multicast or unicast with HER (Head-End Replication) is used to flood BUM (broadcast, unknown destination address, multicast) traffic. RFC 7348, which is incorporated herein by reference as background material.

[0045] The following are examples of the ACI fabric's virtual constructs: tenant, virtual routing and forwarding (VRF) and bridge domain (BD). Current netflow standards are at a switch-level and hence do not augur well for collection of fabric- wide statistics such as an amount of HTTPS traffic on a given tenant, number of packets originating from IP <A.B.C.D> on a given VRF, number of incoming packets across all interfaces in a given BD or traffic originating from a particular distributed application. [0046] One example way to achieve a desired level of fabric-wide aggregation is presented with reference to FIG. 4. FIG. 4 shows an example fabric 400 with various components shown to highlight the process disclosed herein. In one aspect, the concept includes distributing netflow packet collection across all switches in the ACI fabric, i.e, the switches in the fabric can each act as individual netflow collectors. In one example, a source switch (Leaf 1 , Collector 1 (404)) of the incoming packet will calculate a hash on each possible combination of the 5-tuple IP Packet header, along with the tenant, VRF, application and/or interface information. For example, a hash on combinations of each of {tenant, VRF, application, interface} on one side and {SIP, DIP, SPort, DPort, Protocol} on the other side. An example equation could perform this evaluation can include:

[0047] The above example equation computes 7 different combinations of {tenant, VRF, interface} (for x = 3) combined along with 31 different combination of 5-tuple, giving as result of a total of (7 x 31) = 217 possible combinations.

[0048] Each of the resultant hash combinations can correspond to one or more target switches (406, 408, 410, 414 through cloud 412) that will act as netflow collectors for that unique hash. Hence, all the above 217 combinations of netflow packet records will be sent to various destination switches using the underlay network. Each subflow is resident on multiple switches for resiliency and for providing scalable queries.

[0049] Based on the resultant hash values, the Leaf 1 will compute a VxLAN multicast address group and the Leaf 1 will multicast the flow to n leafs. As is shown in FIG. 3, the fi. tenant subflow is sent to Leaf 2 (Collector 2) 406, fi.tenant.wf subflow is sent to Leaf 3 (Collector 3) 408, and the fi.tenant.a _PP subflow is sent through the cloud 412 to a remote vleaf (Collector n+1) 414. [0050] Consider the fi.tenant subflow as an example. The incoming sub-flow to Leaf 2 (Collector 2) 406 will be filtered based on the hash. If the hash is owned by the respective collector, the sub-flow is accepted and saved in the local switch collector database. If not, the sub-flow is rejected. An application policy infrastructure controller (APIC) 416 can use the same hashing algorithm to query any relevant aggregated or granular flow.

[0051] As noted above, computing the VxLAN multicast address group can be performed by a function that maps the sub-flow hash to an IP Multicast address, VxLAN Segment ID combination in the overlay network. A group of leaf switches which share one or more hashes can be considered to be part of a multicast group. A packet corresponding to a hash will be sent on the corresponding multicast group and reaches all the component switches. If a particular switch owns the hash, as in the Leaf 2 406 discussed above, it accepts the packet and creates the flow record in the collector. If a switch doesn't own the hash, it rejects the packet. VxLAN segment ID is used to convey the hash value. The multicast group or multicast rules can be established or set up based on the all possible calculated hash values or hash combination values.

[0052] Note that the hash-to-switch mapping can be done in a distributed fashion in individual switches as it can be computed based on the vector of cluster nodes. Similarly, the hash to multicast group, VxLAN mapping can also be done in a distributed fashion. A switch can automatically join/leave a multicast group based on the hash ownership. In the incoming switch, the packet can be automatically mapped to the multicast group likewise. The above approach makes dynamic addition and deletion of collection nodes automatic and simple. The approach can be particularly useful in a cloud setting, when the collection service is treated as a workload which can be migrated to any node based on the immediate availability of the compute resource. In an ACI fabric distributed across several physical and virtual PODs (a grouping of one or more application images, with additional metadata applied to the POD as a whole), the system could use an orchestrator to place the physical collector instance in any available node. When a new node is added and the hash vector computed, the new node can automatically become part of the collection domain.

[0053] The target switch, say Leaf 2 (406), on receipt of the netflow packet will save the information in the switch's local database, thereby truly achieving distributed netflow collection.

[0054] The collected data can now be queried using the same formula by an administrator, where the query for a particular combination can be distributed to different switches based on the same hash calculation, and then aggregated at the APIC controller 416 level. For example, a query on {tenant A, Protocol X} will correspond to a unique hash, whose results can be queried from the corresponding switch acting as a collector for that combination. The query, in other words, can use the same hashing mechanism used in association with the multicast rules to enable the query to pull the desired data from the fabric.

[0055] The end result is that the approach disclosed herein achieves scalable, distributed netflow along with adding the capability to provide fabric-level netflow statistics even on virtual constructs like tenant, VRF etc.

[0056] Advantages of the disclosed approach include allowing the system to analyze statistics at a fabric-level (as opposed to individual switch- level) as well as enabling the distribution of netflow record collection across the entire fabric, thereby removing a single point of failure, and also enabling us to make distributed queries and data- aggregation. The concept can be extended to the cloud where an Application Virtual Switch (AVS) residing in the cloud can potentially use the ACI fabric as a distributed netflow collector for the stats that the AVS collects. Conversely, the AVS switch itself could act as one of the distributed collectors across the cloud, thereby taking netflow beyond traditional data-center boundaries.

[0057] The disclosed idea provides granular statistics in a distributed fashion, which are otherwise difficult to sustain with static collector configuration. The idea also proposes newer netflow statistics which are more application centric and makes use of ACI constructs. The idea makes use of ACI Leaf and Spine resources for providing collector functionality and uses the fabric multicast service to efficiently deliver the information to the collector. Use of the concept is easily detectible as the solution of fabric wide netflow with a virtual collector needs to be a documented functionality with specific user configuration and guideline. Also, the use of the idea can be viewed by observing the netflow packets on the wire which carry the VxLAN Header and terminating in the Fabric Leaf switches. Collectors can be different leafs in the fabric 400 or can be any other component or node within the fabric.

[0058] FIG. 5 illustrates a method embodiment of this disclosure. A method includes distributing netflow packet collectors across all switched in a network fabric of switches to yield a virtual netflow collector (502), calculating a first hash at a source switch of an incoming packet on each possible combination of a 5-tuple IP packet header (504), calculating a second hash at the source switch of an incoming packet on each possible combination of at least one virtual attribute of the network fabric of switches (506) and creating an N-Tuple flow from the first hash and the second hash (508).

[0059] The method further includes exporting the N-Tuple flow to the virtual netflow collector (510) and mapping, via a virtual extensible local area network multicast address group, the virtual netflow collector to one or more physical netflow collector (512).

[0060] In one aspect, each switch in the network fabric of switches can act as a respective netflow collector. The method can further include receiving a query at the virtual netflow collector regarding the N-Tuple flow (514) and presenting a response to the query based on the mapping of the virtual netflow collector to the one or more physical netflow collector (516). The response can include a visualization response. The query can utilize at least one of the first hash and the second hash or a combined hash to basically use the same hashing algorithm to query as was used to process the flow in the first instance.

[0061] The method can include aggregating statistics for a given virtual attribute to end up in a same physical collector instance according to the mapping of the virtual netflow collector to the one or more physical netflow collector. At least one virtual attribute can include one or more of a tenant, a virtual routing and forwarding object, an endpoint group, a bridge domain, a subnet, a contract, an application, and a filter.

[0062] FIG. 6 illustrates another method embodiment. As shown in FIG. 6, a method includes calculating, at a collector receiving a data flow and via a hashing algorithm, all possible hashes associated with at least one virtual attribute associated with the data flow to yield resultant hash values (602), and based on the resultant hash values, computing a multicast address group (604) and multicasting the data flow to n leafs based on the multicast address group (606). The method includes, at respective other collectors, filtering received sub-flows of the data flow based on the resultant hashes (608) wherein if a respective collector is owned by a hash, the respective collector accepts and saves the sub-flow in a local switch collector database (610). The method also includes receiving a query using the hashing algorithm to query a relevant aggregated or granular flow (612). The system will respond by distributing the query to different switches based on the hash algorithm such that the user can receive fabric- level network statistics even on subflows based on a virtual construct like tenant, VRF, etc.

[0063] In some embodiments the computer-readable storage devices, mediums, and/or memories can include a cable or wireless signal containing a bit stream and the like. However, when mentioned, non-transitory computer-readable storage media expressly exclude media such as energy, carrier signals, electromagnetic waves, and signals per se.

[0064] Methods according to the above-described examples can be implemented using computer-executable instructions that are stored or otherwise available from computer readable media. Such instructions can comprise, for example, instructions and data which cause or otherwise configure a general purpose computer, special purpose computer, or special purpose processing device to perform a certain function or group of functions. Portions of computer resources used can be accessible over a network. The computer executable instructions may be, for example, binaries, intermediate format instructions such as assembly language, firmware, or source code. Examples of computer-readable media that may be used to store instructions, information used, and/or information created during methods according to described examples include magnetic or optical disks, flash memory, USB devices provided with non- volatile memory, networked storage devices, and so on.

[0065] Devices implementing methods according to these disclosures can comprise hardware, firmware and/or software, and can take any of a variety of form factors. Typical examples of such form factors include laptops, smart phones, small form factor personal computers, personal digital assistants, rackmount devices, standalone devices, and so on. Functionality described herein also can be embodied in peripherals or add-in cards. Such functionality can also be implemented on a circuit board among different chips or different processes executing in a single device, by way of further example.

[0066] The instructions, media for conveying such instructions, computing resources for executing them, and other structures for supporting such computing resources are means for providing the functions described in these disclosures.

[0067] Although a variety of examples and other information was used to explain aspects within the scope of the appended claims, no limitation of the claims should be implied based on particular features or arrangements in such examples, as one of ordinary skill would be able to use these examples to derive a wide variety of implementations. Further and although some subject matter may have been described in language specific to examples of structural features and/or method steps, it is to be understood that the subject matter defined in the appended claims is not necessarily limited to these described features or acts. For example, such functionality can be distributed differently or performed in components other than those identified herein. Rather, the described features and steps are disclosed as examples of components of systems and methods within the scope of the appended claims. Moreover, claim language reciting "at least one of" a set indicates that one member of the set or multiple members of the set satisfy the claim.

[0068] It should be understood that features or configurations herein with reference to one embodiment or example can be implemented in, or combined with, other embodiments or examples herein. That is, terms such as "embodiment", "variation", "aspect", "example", "configuration", "implementation", "case", and any other terms which may connote an embodiment, as used herein to describe specific features or configurations, are not intended to limit any of the associated features or

configurations to a specific or separate embodiment or embodiments, and should not be interpreted to suggest that such features or configurations cannot be combined with features or configurations described with reference to other embodiments, variations, aspects, examples, configurations, implementations, cases, and so forth. In other words, features described herein with reference to a specific example (e.g., embodiment, variation, aspect, configuration, implementation, case, etc.) can be combined with features described with reference to another example. Precisely, one of ordinary skill in the art will readily recognize that the various embodiments or examples described herein, and their associated features, can be combined with each other. For example, while some specific protocols such as 802.11 and 802.3 are mentioned in the examples above, the principles could apply to any communication protocol and does not have to be limited to these particular protocols. Any configuration in which received data is acknowledged through an ACK signal could implement the concepts disclosed herein.

[0069] A phrase such as an "aspect" does not imply that such aspect is essential to the subject technology or that such aspect applies to all configurations of the subject technology. A disclosure relating to an aspect may apply to all configurations, or one or more configurations. A phrase such as an aspect may refer to one or more aspects and vice versa. A phrase such as a "configuration" does not imply that such configuration is essential to the subject technology or that such configuration applies to all configurations of the subject technology. A disclosure relating to a

configuration may apply to all configurations, or one or more configurations. A phrase such as a configuration may refer to one or more configurations and vice versa. The word "exemplary" is used herein to mean "serving as an example or illustration." Any aspect or design described herein as "exemplary" is not necessarily to be construed as preferred or advantageous over other aspects or designs.

[0070] Moreover, claim language reciting "at least one of" a set indicates that one member of the set or multiple members of the set satisfy the claim. For example, claim language reciting "at least one of A, B, and C" or "at least one of A, B, or C" means A alone, B alone, C alone, A and B together, A and C together, B and C together, or A, B and C together.