BACKGROUND

[0001] Various types of network switches (e.g., large Ethernet switches) Include several slots which can be populated with line cards. These slots can be populated in various combinations. Line cards can be inserted or removed at run time and any line card can fail at any time. Network operators/administrators are typically responsible for maintaining network operabiiity in view of a line card failure, addition, and/or removal.

[0002] The following description includes discussion of figures having illustrations given by way of example of implementations of embodiments of the invention.

[0003] Figure 1 is a block diagram illustrating a network device according to various embodiments.

[0004] Figure 2 is a block diagram illustrating a network device according to various embodiments.

[0005] Figure 3 is a block diagram illustrating a network device according to various embodiments.

[0006] Figure 4 is a flow diagram of operation in a system according to various embodiments. DETAILED DESCRIPTION

[0007] Network devices with multiple slots available for population by line cards may employ the use of a number of centralized packet switch chips (e.g., located on fabric cards or mounted on the backplane of the network device). Each port on a switch chip may be hardwired to one of the slots on the network device. As such, ports on the switch chips are dedicated to respective slots and the switch chips can turn on and off those ports depending on whether a particular slot is populated with a

functioning/active line card or not. As used herein, a line card refers to one or more electronic circuits on a printed circuit board that interface telecommunication lines coming from subscribers to other parts of a telecommunications network.

[0008] In general, a switch refers to a network bridge that processes and routes data at the data link layer (layer 2) of the OSI (Open Systems Interconnect) model. Switches that additionally process data at the network layer (layer 3 of the OS! model) are frequently referred to as Layer 3 switches, routers or multilayer switches. As used herein, a switch may refer to a Layer 2 bridge or a Layer 3 router. A network switch, as used herein, refers more generally to a networking device that connects network segments.

[0009] The centralized packet switch chips described above may use packet switching (e.g., as opposed to circuit switching) to facilitate the use of packets with the network device. Such switch chips are often expensive and are maximally utilized when all slots are populated with line cards. Thus, the purchaser of a partially populated switch pays for functionality that may never be used.

[0010] Embodiments described herein incorporate packet switching

functionality directly on the line cards within a network device. Packet switching is then implemented within the context of a dynamically reconfigurable optical circuit switching framework.

[0011] Figure 1 is a block diagram illustrating a network device according to various embodiments. As shown, network device 100 includes a centralized circuit switch 1 10 connected to a plurality of line cards 120- 150. The number of line cards illustrated in Fig, 1 is merely an example of various embodiments: more line cards or fewer line cards could be employed in different embodiments. It should also be noted that the number of line cards illustrated does not necessarily equal the number of line card slots on network device 100. in other words, the four illustrated line cards may only occupy a subset of available slots in network device 100. In general, network device 100 may be an Ethernet switch or other suitable device for routing packet traffic

[0012] Line cards 120-150 each have a packet switching module 122- 152, respectively. Packet switching modules 122-152 can be

implemented as one or more hardware modules, special-purpose hardware (e.g., application specific hardware, application specific integrated circuits (ASICs), embedded controllers, hardwired circuitry, etc.), software modules or some combination of these. As the name implies, packet switching modules 122-152 perform packet switching. As used herein, packet switching refers to delivery of variable-bit-rate data streams (i.e., sequences of packets) over a network. In various embodiments, each packet switching module is capable of routing packets on behalf of other packet switching modules on network device 100. For example, packet switching modules 122-152 could each route packets using a universal routing scheme such as the routing scheme proposed by Valiant (L G. Valiant A scheme for fast parallel

communication. SI AM Journal on Computing, 11(2}:350-361 _! 1982).

[0013] Each of line cards 120-150 is connected to circuit switch 1 10. As shown, each line card includes multiple links (e.g., bi-directional links) to circuit switch 1 10. While the specific number of links from each line card may vary, various embodiments include at least a plurality of links to provide a level of redundancy.

[0014] In various embodiments, circuit switch 1 10 is an optical circuit switch. In general, a circuit switch refers to a switch that establishes a circuit or channel between network nodes, as if the nodes were physically connected with an electrical circuit. Accordingly, an optical circuit switch, as used herein, refers a circuit switch that uses optical signals and links for communication. Below is an example of how an optical circuit switch might be constructed. Other suitable optical circuit switch(es) and/or optical switching framework could be used in various embodiments described herein. [0015] Waveguides (e.g., hollow metal, polymer core, etc.) may be formed in a backplane to create an optical crossbar framework. For example, a first group of parallel crossbars may be bus lines which traverse the length of the backplane, while a second group of parallel crossbars may be tap lines which intersect the bus lines and are connected to a computing element.

[0016] An optical element or combination of optical elements may be placed in each intersection to selectively direct optical signals from the bus lines into the tap lines and vice versa. These optical elements may include prisms, mirrors, light valves and other optical elements. The optical elements may be dynamic or passive. At least one of the optical elements in the intersection may change state to switch the

interconnection from a through state to a crossed state.

[0017] Computing elements are connected to the optical crossbar

framework. For example, a primary computing device or network may be connected to the bus lines and a number of other computing devices may be connected to the tap lines. Bus lines may be connected to a larger computing network or router and the tap lines may be connected to a number of blade computer elements. Each of the blade computer elements may be connected to multiple tap lines. Computing devices may use more or less tap lines for bidirectional communication.

Additionally, the computing device may use wavelength division multiplexing for bidirectional communication over a given line set. [0018] As discussed above, the bus ports may be connected in a variety of locations along the bus lines. In some embodiments, it may be advantageous to locate the bus ports at one end oi the bus lines. In other embodiments, buses ports may be attached to the center of the bus lines and distribute optical signals in both directions through the bus lines. In general, the location of the bus ports on the bus lines can be determined by a number of factors including: space constraints, connection constraints, optical loss budgeting, or other relevant criteria.

[0019] Optical elements can be dynamically switched within the optical crossbar framework to connect desired tap lines to the bus lines. For example, dynamically switching the optical elements may include separately switching a number individual elements or moving a block of elements with a single actuator. By way of example and not limitation, this switching may include moving a solid periscope prism into the path of a bus line, tilting mirror into the path of a bus line, or opening a light valve to allow passage of optical signals from a bus line into a tap line.

[0020] Optical signals are directed between the tap lines and bus lines.

For example, the optical signal from multiple tap lines may be connected to a single bus line using a series of combiner elements. Additionally or alternatively, a single tap line may be simultaneously connected to two or more bus lines.

[0021] In some embodiments, optical circuit switch 1 10 is a single

centralized circuit switch, in other embodiments, optical circuit switch 1 10 is comprised of several smaller circuit switches connected via optical links. In such embodiments, circuit switch 100 may be distributed, for example, among active line cards (see, e.g., Fig. 3). in other words, in addition to each line card having a packet switching module, each line card (or at least some of the line cards) may have a circuit switching module. Distributed circuit switching modules may be controlled to act in concert as a single circuit switch, as described in more detail below.

[0022] Switching manager 160 adaptiveiy reconfigures optical circuit switch 1 10 based on the configuration of line cards on network device 100. in other words, switching manager monitors line card slots on network device 100 and determines which slots are populated with active line cards. Based on the active line card configuration, switching manager 180 adaptiveiy configures circuit switch 1 10 to scale packet switching bandwidth accordingly. Switching manager 160 can be implemented as one or more hardware modules, special-purpose hardware (e.g., application specific hardware, application specific integrated circuits (ASICs), embedded controllers, hardwired circuitry, etc.), software modules or some combination of these.

[0023] At the direction of switching manager 160, optical circuit switch 1 10 (as opposed to, for example packet switching modules 122-152) handles link redundancies and/or link failures for network device 100. For example, if a link fails, switching manager 160 reconfigures optical circuit switch 1 10 to compensate for the link failure and maintain maximum bandwidth on the remaining active links. While switching manager 160 directs, calculates and/or determines the reconfiguration of optica! circuit switch 1 10, it is optical circuit switch 1 10 that ultimately carries out the

[0024] Figure 2 is a block diagram illustrating a system according to various embodiments. Similar to Fig. 1 , network device 200 includes various removable line cards 220-250, a optical backplane 270, an optical circuit switch 210, and a switching manager 260. As with embodiments described above, circuit switch 210 may be an optical circuit switch that is distributed, for example, among line cards via optical links.

[0025] As shown, line cards 220-250 are connected to optical circuit switch 210 via optical backplane 270. In various embodiments, optical backplane 270 may be a passive optical backplane, which offers no active bus driving circuitry but is not considered a single point of failure for on the network.

[0026] Optical circuit switch 210 is configurable (e.g., via switching

manager 260) to connect one line card to another line card at a full- bandwidth rate (perhaps via intermediate switches). With conventional, centralized packet switching, centralized packet switches may be over- provisioned to maintain full-bandwidth connections. By establishing the optical circuit switching framework of circuit switch 210, packet switching modules 222-252 can be guaranteed full-bandwidth connections with normal provisioning.

[0027] Additionally, centralized packet switches are often hardwired to each line card. Conversely, in various embodiments described herein, if one of line cards 220-250 is removed, switching manager 260

reconfigures the circuit switching framework of circuit switch 210 to maintain guaranteed packet switching bandwidth between the remaining active line cards. Circuit switch 210 also handles link failure and link redundancy via switching manager 260. in various embodiments, switching manager 260 is implemented as a management processor. In other embodiments, switching manager 260 may be implemented as one or more hardware modules, special-purpose hardware (e.g., application specific hardware, application specific integrated circuits (ASICs), embedded controllers, hardwired circuitry, etc.), software modules or some combination of these.

[0028] Figure 3 is a block diagram illustrating a network device according to various embodiments. In particular, network device 300 illustrates a distributed optical circuit switch. The distributed optical circuit switch is at least distributed across line cards 320-350, which are all optically connected to optical backplane 370. in certain embodiments, the distributed circuit switch may additionally include independent circuit switch 380 and/or circuit switch 381 . In other words, various combination of circuit switching modules on line cards and independent circuit switches can be combined to form an overall optical circuit switch. The optical circuit switching functionality is controlled and managed by switching manager 360. If one or more line cards fail, a particular link fails, or other event occurs, switching manager 360 reconfigures the overall optical circuit switch by coordinating changes to the various optica! circuit switching modules 323-353, circuit switches 380-381 , and/or optica! backplane 370. As with other embodiments, switching manager 380 may be implemented as one or more hardware modules, special-purpose hardware (e.g., application specific hardware, application specific integrated circuits (ASICs), embedded controllers, hardwired circuitry, etc.), software modules or some combination of these.

[0029] Figure 4 is a flow diagram of operation in a system according to various embodiments. Fig. 4 includes particular operations and execution order according to certain embodiments. However, in different embodiments, other operations, omitting one or more of the depicted operations, and/or proceeding in other orders of execution may also be used according to teachings described herein.

[0030] Initial switching resources are determined 410. For example, a switching manager on a network device may determine which slots have active line cards, which line cards have optical circuit switching modules and any other switching resources (e.g., packet switching modules) on the network device. Based on the available resources, an optical circuit switching framework is established 410 tor the device. In particular, the optical circuit switching framework is established on the network device in view of the active (and removable) line cards installed in the network device. The optical circuit switching framework is configured to maximize packet switching functionality (e.g., packet switching modules) included on each of the active line cards. [0031] Accordingly, traffic is routed 430 through the network device via packet switching on the removable line cards. Packet switching is performed within the bounds of the opticai circuit switching framework, in other words, the opticai circuit switching framework defines (at least logically) the network topology of the network device. Given that topology, packet switches on the line cards operate to route packets via packet switching in view of the topology.

[0032] When a link status event (e.g., a link failure, line card insertion, line card removal, etc.) is detected, the event is automatically resolved. For example, if the configuration of the line cards changes, the opticai circuit switching framework is automatically reconfigured 440 (e.g., via a switching manager) to maintain link redundancy and guaranteed packet switching bandwidth among active line cards. The guaranteed packet switching bandwidth may be a fuil-bandwidth guarantee or a guarantee of something less than fuil-bandwidth.

[0033] Various components, modules, etc. described herein may be a means for performing associated functions, operations, etc.