Switching network module and switching network

Abstract

A switched network module and a switched network are provided. A switched network network module has a plurality of multi-fiber connector interfaces, where some of these multi-fiber connector interfaces may be connected to network devices in a network having an internal mesh implemented in two levels using multi-fiber cables. The first hierarchy is configured for rearrangement, and the second hierarchy is configured for recombining individual optical fibers of different groups of optical fibers. The optical path of each transmitter and receiver is matched so as to provide an appropriate optical connection from the transmitting fiber to the receiving fiber, and any complex network topology may be achieved by at least a reduced 1 / N point-to-point interconnection, where N = the number of channels per multi-fiber connector interface.

Description

Technical Field

[0001] Disclosed are a switching network module device and method for improving the scalability of data center networks using a mesh network topology and switches with various cardinalities, hierarchies, and oversubscription ratios. The disclosed switching network module device and method reduce the number of manual network connections, simplify cabling installation, and improve data center flexibility and reliability. Background Art

[0002] The importance of optical fiber for transmitting communications signals is rapidly growing due to its high bandwidth, low attenuation, and other significant advantages, including radiation resistance, small size, and light weight. Data center architectures using optical fiber are evolving to meet global traffic demands and the ever-increasing number of users and applications. The rise of cloud data centers, particularly hyperscale clouds, has significantly changed enterprise information technology (IT) business structures, network systems, and topologies. Furthermore, the requirements of cloud data centers are influencing technology roadmaps and standardization.

[0003] The widespread adoption of server virtualization and advances in data processing and storage technologies have led to a surge in east-west traffic within data centers. Traditional three-tier switch architectures, comprising core, aggregation, and access layers (CAA), cannot provide the low and balanced latency paths required for east-west traffic. Furthermore, because CAA architectures utilize the spanning tree protocol to disable redundant paths and build a loop-free topology, they underutilize network capacity.

[0004] The Folded Clos network (FCN) or Spine-and-Leaf architecture is a topology that is more suitable for overcoming the limitations of three-tier CAA networks. The Clos network is a multi-stage circuit-switched network proposed by Charles Clos in 1953. Initially, the network was designed to increase the capacity of crossbar switches. Due to the development and adoption of very large-scale integrated circuit (VLSI) technology, the Clos network became less important. Complex optical interconnect topologies were initially used in high-performance computing (HPC) and later in cloud data centers, making this architecture important again. The Folded Clos network topology utilizes two types of switching nodes: spine and leaf. Each spine is connected to every leaf. The network can be scaled horizontally to enable communication between a large number of servers while minimizing latency and unevenness by simply adding more spine and leaf switches.

[0005] An FCN depends on the switch radix k, which is the ratio of server downlinks from leaf switches to uplinks from spine switches, and the number of tiers or layers, m, in the network. The choice of (k, m) has a significant impact on the number of switches, the reliability and latency of the network, and the deployment cost of the data center network. Figure 1 Shown is the relationship between the number of servers for different levels of oversubscription, assuming all switches have similar cardinality and a total oversubscription ratio of 1:1.

[0006] Figure 2 An example of two FCNs with a similar number of hosts using different radixes and levels is shown. The higher radix (32 in this example) connects 32 edge switches in a two-tier network, as shown in part (a) of the figure. A two-tier FCN provides the lowest latency, but at the expense of a dense network (512 interconnects). By using a three-tier network, the interconnect layout is simplified (256 interconnects). However, this requires more switches and introduces more latency into the network. Over the past few years, the need for flatter networks to handle the growing amount of traffic between machines has led to an increase in the radix of switch application-specific integrated circuits (ASICs). Currently, ASICs can handle 256-radix switches at 100Gb / s per port. These switches support 64x400GbE, 128x200GbE, or 256x100GbE, supporting flat networks with up to three tiers.

[0007] Based on the industry telecommunications infrastructure standard TIA-942-A, leaf and spine switches can be located tens or hundreds of meters apart. Typically, spine switches are located in the main distribution area (MDA), while leaf switches are located in the equipment distribution area (EDA) or horizontal distribution area (HDA).

[0008] This architecture has been proven to deliver high bandwidth and low latency (only two hops to reach the destination) with low oversubscription connectivity. However, for a large number of switches, the spine-and-leaf architecture requires a complex mesh with a large number of fibers and connectors, which increases the cost and complexity of installation.

[0009] Future data centers will require more flexible and adaptable networks than the traditional grids implemented today to accommodate highly distributed computing, machine learning (ML) training workloads, advanced virtualization, and data replication.

[0010] Deploying a new data center or expanding a data center network with hundreds or thousands of servers is not an easy task. Spine-to-leaf switches require a large number of interconnects, such as Figure 3In this example, the fabric 100 may have 572 paths. Each line in illustration 120 may represent a group of eight or twelve optical fibers terminated in a multi-fiber MPO connector. The optical fibers may be ribbonized in a conventional flat ribbon or a rollable ribbon format. Illustration 110 shows a magnified view of a small area of ​​the fabric 100.

[0011] Interconnected switches with 100 or more switches can be prone to errors, which, in many cases, can be exacerbated by tight deployment deadlines or lack of installation personnel training. While spine-and-leaf topologies are resilient to misaligned connections, large interconnect errors can have a noticeable impact through performance degradation, leading to the loss of some server links. Managing large-scale network configurations often requires dedicated personnel to check interconnects, which can cause delays and increase deployment costs.

[0012] As shown in the prior art, the use of transpose boxes can help reduce installation errors. However, the prior art cannot easily adapt to different network topologies, switch bases, or oversubscription levels.

[0013] Disclosed herein is a new grid method and apparatus that utilizes a modular, flexible, and better organized interconnect mapping that can be quickly and reliably deployed in a data center.

[0014] US8621111, US2012 / 0250679A1, and US2014 / 0025843A1 disclose a method for providing scalability in a data transmission network using a transpose box. This box can connect the first and second layers of a network. This box facilitates network deployment, but requires a dedicated box for the selected network. As described in these applications, the network topology determines the type of transpose box to use. Changes in topology may require replacing the transpose box. According to the specifications, a different box may be required if the number of spine or leaf switches changes, if there is oversubscription, or if other parameters of the network change.

[0015] Once a topology is selected, the application provides a method for expansion. This requires connecting a port on one box to a port on another box with a cable. This increases network loss and does not effectively accommodate network expansion.

[0016] The approach disclosed in US 2014 / 0025843 A1 may be well suited for large data centers that have already chosen the type of network architecture to implement and can prepare and maintain an inventory of different types of transpose boxes based on their needs. Wider deployment of mesh networks in data centers requires a more flexible or modular approach.

[0017] In W2019099771A1, an interconnect box is disclosed. This application shows an exemplary cable connecting various spine and leaf switches using a 1RU module that can be mounted on a rack. The ports of these modules are internally connected using an internal multi-fiber cable that contains a specific mesh. However, the module appears to be tuned to a specific topology, such as providing a mesh between four spine and leaf switch ports. The application does not describe how the device can be used with a topology with a variable number of leaf switches or spine switches or a variable number of ports.

[0018] US20150295655A describes an optical interconnect assembly that uses multiple leaf-side multiplexers and demultiplexers on each side of the network, one located on the spine and another located near the leaf. Each multiplexer and demultiplexer is configured to work together in a desired topology. However, the application does not demonstrate the flexibility and scalability of this approach.

[0019] US11269152 describes a method for circumventing the limitations of optical shuffle boxes, which, according to the application, are not easily adaptable to reconfiguration or expansion of a switch network. The application describes an apparatus and method for patching a network link using multiple patch panels. At least two racks are required to connect a switch from one layer of the network to another. Each rack is capable of accommodating multiple modules, such as boxes arranged in a vertical configuration. The connection from the first-tier switch to one side of the module is made using a branch cable. One side of the branch cable terminates with an MPO (24 fibers) and the other side terminates with an LC or other duplex connector. One side of the module has one or two MPO ports, and an additional six duplex LC connectors or newer ultra-small form factor (VSFF) connectors.

[0020] Similarly, the second-tier switches are connected to modules in other racks. The wiring required to connect the switches is performed using a plurality of patch cord assemblies configured to connect to a plurality of optical modules. The patch cords are specially designed to fix their relative positions as they must maintain the correct (linear) sequence. US11269152 describes a patching method that can make the network more scalable based on the network base. However, network deployment remains challenging and prone to interconnection errors. Utility Model Content

[0021] A device having multiple multi-fiber connector interfaces, some of which are capable of connecting to network devices in a network using multi-fiber cables, the network having an internal grid implemented in two levels. The first level is configured for rearrangement, and the second level is configured for recombination of individual fibers from different fiber groups. The optical paths of each transmitter and receiver are matched to provide appropriate optical connections from the transmitting fiber to the receiving fiber, and complex arbitrary network topologies can be implemented by reducing the number of point-to-point interconnects by at least 1 / N, where N = the number of channels per multi-fiber connector interface. Moreover, the fiber interconnects within the device can transmit signals at any wavelength used by the transceiver (e.g., 850nm-1600nm). Due to the transparency of the fiber interconnects in the device, signals of each wavelength can be assigned to unidirectional propagation or bidirectional propagation from the transmitter to the receiver. BRIEF DESCRIPTION OF THE DRAWINGS

[0022] Figure 1 The number of servers is shown as a function of the switch radix and the number of switch layers in the network.

[0023] Figure 2 An example of two FCNs with a similar number of hosts using different cardinalities and levels is shown.

[0024] Figure 3 The interconnection of an example grid containing 576 interconnects (each having 12 or 8 optical fibers) is shown.

[0025] Figure 4 (a) shows a front view of the disclosed module 400.

[0026] Figure 4 (b) shows a rear view of the module 400 .

[0027] Figure 5 A top view of module 400 is shown.

[0028] Figure 6 The interconnection of modules 400 is shown.

[0029] Figure 7 The interconnection of regions 480 of module 400 is shown.

[0030] Figure 8 is a top view of the submodule 500 showing the interconnection arrangement.

[0031] Figure 9 Shown are 16 possible configurations that can be implemented in the submodules.

[0032] Figure 10 The diagram illustrates a simple method of using module 400 to implement a network with 16 leaf switches and up to 16 spine switches.

[0033] Figure 11 (a) shows an example of the interconnection between the spine port and the module 400.

[0034] Figure 11 (b) shows Figure 11 (a) Example interconnection table.

[0035] Figure 12 (a) shows an example of the interconnection between the ports of the module 400 and the spine chassis ports (eight spines with two line cards per spine).

[0036] Figure 12 (b) shows an example of the interconnection between the ports of the module 400 and the spine chassis ports (eight spines, with four spines having four line cards each).

[0037] Figure 12 (c) shows an example of the interconnection between the ports of the module 400 and the spine chassis ports (eight spines, with two spines having eight line cards each).

[0038] Figure 13 A method for implementing a two-level FCN is shown.

[0039] Figure 14 A method for implementing a two-level FCN is shown.

[0040] Figure 15 A method for implementing a two-level FCN is shown.

[0041] Figure 16 A method for implementing a two-level FCN is shown.

[0042] Figure 17 A method for implementing a two-level FCN is shown.

[0043] Figure 18 A method for implementing a three-level FCN is shown.

[0044] Figure 19 A method for implementing a three-level FCN is shown.

[0045] Figure 20 The figure shows a 3-tier FCN deployment with 32 PODs and 16 switches, each POD using a stack of modules 400. The figure shows the front side of the stack. L (abbreviation for leaf) and p (POD) represent leaf switches and POD numbers, respectively.

[0046] Figure 21The figure shows a 3-tier FCN deployment with 32 pods and 16 switches, each pod using a stack of modules 400. The figure shows the back side of the stack. S (spine) and c (line card) represent the spine switch and line card number, respectively. DETAILED DESCRIPTION

[0047] This document discloses a modular apparatus and general method for deploying optical networks of various tiers and radixes. The modules and method can be used with standalone, stacked, or rack network switches, as long as the modular connections utilize MPO connectors with eight or more fibers. In particular, switches with Ethernet-specified SR or DR transceivers in the module ports (such as 40GBASE-SR4, 100GBASE-SR4, 200GBASE-SR4, or 400GBASE-DR4) can use these modules without changing connectivity. Networks with single-channel duplex transceivers (10G SR / LR, 25G SR / LR), (100GBASE-LR4, 400GBASE-LR4 / FR4) will also work with these mesh modules, provided that the correct TX / RX polarity is maintained in the mesh. Other types of transceivers (such as 400GBASE-FR4 / LR4) can also be used by combining the four transceiver ports with a harness or breakout box.

[0048] Figure 4 (a) shows a front view of the disclosed module 400, which is a key element to facilitate the deployment, remodeling and expansion of optical networks. In this embodiment, the module has 32 MPO connector ports, which can be divided into a front part and a rear part, such as Figure 4 Alternatively, the 32 ports may be located on one side of the device (not shown here).

[0049] For ease of illustration, we assume that ports 420 to 435 (each with four MPO connectors, labeled a, b, c, and d) are located on the front side of the module, facing the leaf switch, as shown in portion (a) of the figure. On the other side of the module, ports 440 to 470 (opposite ports 420-435), each representing an MPO connector, face the spine switch connections. The MPO dimensions allow the module width W to range from 12 inches to 19 inches and the height H to range from 0.4 inches to 0.64 inches. The small width of the 16 MPO connectors relative to the width of the rack (19 inches) provides ample space for machine-readable labels 410, 412 and visual labels 414, 413, which can aid in deploying or inspecting network interconnects, as described later in this application. Furthermore, if desired, lateral rails 405 on both sides of the module will enable the module to be inserted into the rack structure. Alternatively, the module can be attached directly to the rack using brackets 406. By using the specified height ranges of this embodiment, up to four modules can be stacked in 1RU or less than 1.5RU, depending on density requirements.

[0050] Figure 5 A top view of the module is shown, showing attached machine-readable tags 410 and 412. The tags can be read by a laser scanner or camera. The read code can be linked to a database containing an interconnection diagram of all modules in the network. This information can be displayed on a portable device, tablet, phone, or augmented reality lens to facilitate deployment. For more specific information on this, see RS16563 and 25512.

[0051] Figure 6 The interconnection scheme of the module according to the present invention is shown. The interconnection configuration of all module ports is described in Table I and Table II. In order to simplify the module structure, the grid is divided into two areas, 480 and 490. Area 480 reorders the fiber groups, for example, 482 is paired with 485, and 485 can be a standard or reel ribbon, or just a cable unit with 8, 12 or 16 fibers. The connection method required to simplify the leaf switch connection is as follows Figure 7 As described above, in region 490, the mesh is implemented at the fiber level. For example, fiber 485 from fiber group 420a and fiber 482 from fiber group 435a are intermixed with fibers from the other two groups 425a and 430a. In this embodiment, four submodules 500 are used to create the interconnected mesh of the four fiber groups shown in this embodiment. Figure 8A connection diagram for one of the submodules 500 is shown. In this diagram, we show how the fibers in port groups 510 to 525 are mixed with other fibers in groups 515 and 520 from submodule 480. On the opposite side, depending on the position of submodule 500, outputs 550-565 of submodule 500 can correspond to four module ports (e.g., 440-446). Thus, the apparatus according to the present invention mixes Ethernet physical medium dependent (PMD) channels with other transceiver PMD channels to distribute network data flow and help balance the data flow load to any one transceiver.

[0052] For an MPO transmitting four parallel channels, a grid of submodules 500 can be implemented in a large array arrangement. For an MPO connector with Nf = 12 fibers, Nc = 4 duplex channels, and Np = 4 multi-fiber connector ports, the following equations describe the output port, I A and I B To output port O A and O B The topological mapping preserves the correct path from transmitter to receiver.

[0053] Input port: I A =i+Nf×(k-1),I B =1-i+Nf×k,(1)

[0054] Output port: O A =p(i,r1)+Nf×(p(k,r2)-1),O B =1-p(i,r1)+Nf×p(k,r2),(2)

[0055] In (1) and (2), i is an index from 1 to Nc, associated with the input duplex port of the connector, k is an index of the connector ranging from 1 to Np, and p(.,.) is a permutation function that has two input parameters, the first being the number to be permuted and the second being the order of the permutations in the list of Nc! = 24 possible permutations. These sets of equations indicate that r1 and r2 determine the number of possible configurations; thus, module 500 may have I A Connect to O A And I B Connect to O B There are r1xr2=576 possible configurations, and a total of 1152 possible configurations when using cross-connection (e.g. Figure 9 I is shown in A to O B Table 1 describes their interconnection arrangements, enabling efficient use of connection methods such as TIA 568.3D Method A or Method B.

[0056] 610 612 614 616 618 620 622 624 626 628 630 632 634 636 638 640 510a 550a 550a 550a 565d 565b 565d 565c 565a 550h 550h 550h 565e 565g 565e 565f 565h 510b 555b 555b 560d 560d 560b 560d 560c 560a 555g 555g 560e 560e 560g 560e 560f 560h 510c 560c 560c 550b 555d 555b 555d 555c 555a 560f 560f 550g 555e 555g 555e 555f 555h 510d 565d 565d 565d 550d 550b 550d 550c 550a 565e 565e 565e 550e 550g 550e 550f 550h 510e 565e 565e 565e 550e 550g 550e 550f 550h 565d 565d 565d 550d 550b 550d 550c 550a 510f 560f 560f 550g 555e 555g 555e 555f 555h 560c 560c 550b 555d 555b 555d 555c 555a 510g 555g 555g 560e 560e 560g 560e 560f 560h 555b 555b 560d 560d 560b 560d 560c 560a 510h 550h 550h 550h 565e 565g 565e 565f 565h 550a 550a 550a 565d 565b 565d 565c 565a 515a 555a 555a 555a 565c 565a 565b 565d 565c 555h 555h 555h 565f 565h 565g 565e 565f 515b 550b 550b 560b 560c 560a 560b 560d 560c 550g 550g 560g 560f 560h 560g 560e 560f 515c 560b 560b 555b 555c 555a 555b 555d 555c 560g 560g 555g 555f 555h 555g 555e 555f 515d 565b 565b 565b 550c 550a 550b 550d 550c 565g 565g 565g 550f 550h 550g 550e 550f 515e 565g 565g 565g 550f 550h 550g 550e 550f 565b 565b 565b 550c 550a 550b 550d 550c 515f 560g 560g 555g 555f 555h 555g 555e 555f 560b 560b 555b 555c 555a 555b 555d 555c 515g 550g 550g 560g 560f 560h 560g 560e 560f 550b 550b 560b 560c 560a 560b 560d 560c 515h 555h 555h 555h 565f 565h 565g 565e 565f 555a 555a 555a 565c 565a 565b 565d 565c 520a 560a 560a 560a 565b 565c 565a 565b 565d 560h 560h 560h 565g 565f 565h 565g 565e 520b 550c 565c 560c 560b 560c 560a 560b 560d 550f 565f 560f 560g 560f 560h 560g 560e 520c 555c 555c 555c 555b 555c 555a 555b 555d 555f 555f 555f 555g 555f 555h 555g 555e 520d 565c 550c 550c 550b 550c 550a 550b 550d 565f 550f 550f 550g 550f 550h 550g 550e 520e 565f 550f 550f 550g 550f 550h 550g 550e 565c 550c 550c 550b 550c 550a 550b 550d 520f 555f 555f 555f 555g 555f 555h 555g 555e 555c 555c 555c 555b 555c 555a 555b 555d 520g 550f 565f 560f 560g 560f 560h 560g 560e 550c 565c 560c 560b 560c 560a 560b 560d 520h 560h 560h 560h 565g 565f 565h 565g 565e 560a 560a 560a 565b 565c 565a 565b 565d 525a 565a 565a 565a 565a 565d 565c 565a 565b 565h 565h 565h 565h 565e 565f 565h 565g 525b 560d 560d 565c 560a 560d 560c 560a 560b 560e 560e 565f 560h 560e 560f 560h 560g 525c 555d 555d 555d 555a 555d 555c 555a 555b 555e 555e 555e 555h 555e 555f 555h 555g 525d 550d 550d 550d 550a 550d 550c 550a 550b 550e 550e 550e 550h 550e 550f 550h 550g 525e 550e 550e 550e 550h 550e 550f 550h 550g 550d 550d 550d 550a 550d 550c 550a 550b 525f 555e 555e 555e 555h 555e 555f 555h 555g 555d 555d 555d 555a 555d 555c 555a 555b 525g 560e 560e 565f 560h 560e 560f 560h 560g 560d 560d 565c 560a 560d 560c 560a 560b 525h 565h 565h 565h 565h 565e 565f 565h 565g 565a 565a 565a 565a 565d 565c 565a 565b

[0057] Table 1 Grid configuration table of sixteen possible arrangements of 610 to 640 of submodule 500

[0058] 440 442 444 446 448 450 452 454 456 458 460 462 464 466 468 470 420a 420a 420a 420a 420b 420b 420b 420b 420c 420c 420c 420c 420d 420d 420d 420d 425a 425a 425a 425a 425b 425b 425b 425b 425c 425c 425c 425c 425d 425d 425d 425d 430a 430a 430a 430a 430b 430b 430b 430b 430c 430c 430c 430c 430d 430d 430d 430d 435a 435a 435a 435a 435b 435b 435b 435b 435c 435c 435c 435c 435d 435d 435d 435d

[0059] Table II Grid Configuration of Module 400

[0060] The two-step mesh incorporated into each module 400 by combining sections 480 and 490 increases the mix of fiber channels within each module. This simplifies network deployment because a significant portion of the network's complexity is shifted from the structured cabling switch fabric to one or more modules 400. The fibers in sections 480 and 490 are connected together by 495, which represents a connector or splice. Note that at this connection point, the fiber array in section 480 can be flipped to accommodate different interconnection methods, such as TIA 568.3D Method A or Method B. Using modules 400 and following simple rules to connect a group of uplinks or downlinks horizontally or vertically, installation becomes cleaner and cable management is greatly improved, as shown in the following description of this application.

[0061] A set of N modules 400 can implement different configurations of the radix, with different configurations having different numbers of spine and leaf switches. For example, Figure 10 A stack of four modules 400 is shown. Part (a) of the figure shows the side of the module connected to the leaf switch. For simplicity, we will mark it as the front side. Part (b) of the figure shows the opposite side of the same module 400, the back side, which is connected to the spine switch.

[0062] Figure 10 The figure assumes that 16 leaf switches need to be connected to the switching network shown in part (d) of the figure, and each leaf switch has 4 MPO uplinks. In this illustrative example, the uplinks of each leaf switch are connected horizontally in groups of four until the last port of each module 400 is used. For example, 710 and 712, the first and last fourth ports of the first module 400 are connected to the uplink ports of the leaf switches L1 and L4, respectively. The uplink of the fifth leaf switch fills the port 714 of the second module 400. This method continues until the uplink of the last leaf switch is connected to the port 716.

[0063] The spine ports are assigned on the back of the stacking module 400. For example, if separate spine switches are used, 720, 722, and 724 correspond to the ports of the first, second, and sixteenth spine switches, respectively. Figure 10The connections from the modules to the spine are described in more detail as follows: Figure 11 described.

[0064] Alternatively, spines can be implemented using rack switches. Although more expensive than standalone systems, rack switches can provide several advantages such as scalability, reliability, and performance. The port connections for the spines using rack switches can follow various arrangements. For example, using eight spine switches, each with two line cards, all S1 and S2 ports can be connected to the first spine, S3 and S4 to the second spine, and S15 and S16 ports to the last spine. Using four spine switches, each with four line cards, all S1, S2, S3, and S4 ports can be connected to the first spine, S5, S6, S7, S8 to the second spine, and S13, S14, S15, S16 to the last spine. If only two spine switches, each with eight line cards, are used, then all ports S1, S2, S3 to S8 will be connected to the first spine ( Figure 9 The S1' in the figure, and the S9 to S16 ports will be connected to the second spine (S2'). A more detailed description of the connection from the module to the spine is as follows Figure 12 As described in Tables (a), (b) and (c) in .

[0065] In many cases, such as when using chassis switches with many line cards, the number of spine switches may be less than 16. In these cases, several ports can be grouped to populate the spine switches. For example, 730 groups 32 ports to connect to spine S1', and the other 32 ports labeled 732 connect to the second spine (S2'). By using module 400 and the described method, each spine switch is interconnected with all leaf switches, such as Figure 10 The grid representation shown in 755 can be obtained by following the equation in Figure 750. Figure 7 , as verified by the connection tables in Tables I and II. Generally speaking, module 400 reduces the complexity of scaling up or down networks, as shown. The interconnections within the device can transmit signals at any wavelength in the 830nm-1650nm range. Furthermore, the signal assigned to each wavelength can propagate unidirectionally or bidirectionally from transmitter to receiver. The interconnection ports utilize multi-fiber connectors of varying form factors, such as CS, SN, MPO, SN-MT, and MMC.

[0066] Deployment example of a network using module 400

[0067] Figures 13 to 17The examples in show implementations of two-tier and three-tier FCNs using various cardinalities, oversubscriptions, and sizes of modules 400. Tables III through VI detail the number of modules required for each network and the estimated rack space required for the modules.

[0068] Starting from a two-layer FCN, Figure 13 Two switch fabrics are shown: 810 and 820, each with 16 spine switches. The switch fabric 810 shown in the figure can be implemented using four modules 400. The connection diagram shows a stack of modules 400 from both sides of the module, one side labeled as the front, 815, and the other side labeled as the back, 817. Side 815 is connected to 32 leaf switches with four MPO uplinks and eight fibers allocated for duplex connectivity. These switches are labeled Li, where i is the index of the switch. For this example, i ranges from 1 to 16. As shown on side 815, the leaf switches are connected horizontally. All L1 uplinks are connected adjacently in the first four ports of the first module 400. All L32 uplinks are connected to the last four ports of the eighth module 400. Starting from the back 816 side of the same module stack, the 16 spine switches are connected vertically as shown. Based on the disclosed dimensions of the modules 400, this switch fabric can be implemented in less than 3RU.

[0069] The switch fabric 820 has 64 leaf switches with four MPO uplinks, and four modules 400 can be used to implement the switch fabric 820. The connection method is similar to that described above. From the 825 side, all leaf switch uplinks are connected adjacently following a consistent method. For example, L1 is connected to the first four ports of the first module 400. All L64 uplinks are connected to the last four ports of the sixteenth module 400. Starting from the back 826 side of the same module stack, 16 spine switches are connected vertically as shown. Based on the size of the disclosed module 400, the switch fabric can be implemented in less than 5RU.

[0070] Figure 14 and Figure 15 The network in Figure 8 has the same number of spine switches, but a much larger number of leaf switches. The implementation is similar. For 830, the L1 is connected to the first four ports of the first module 400, and the L64 uplink is connected to the last four ports of the thirty-second module 400. As mentioned above, the spine switches are connected vertically. The network shown in Figure 850 requires 128 modules, and due to the higher number of ports, the spine needs to be implemented using a chassis with 16 line cards.

[0071] The switching fabric described below has radix 32 leaf switches, meaning they have 16 uplinks (4 MPOs) and 16 downlinks (4 MPOs). Figure 16A network using radix 64 leaf switches with 8 MPOs on the uplink and 8 MPOs on the downlink is shown.

[0072] Implementing this network can result in a lower oversubscription ratio, such as 1:1, but at the expense of greater complexity. Module 400 can also be used to simplify installation. Figure 16 As shown, 16 modules are used to connect to 64 leaf switches. The uplinks of all leaf switches are divided into two groups, each with 4 MPOs. This separation does not require any special commands or procedures. For network 820, each group is exactly as Figure 13 This means that using 16 modules 400, the first set of 4 MPO uplinks per leaf grid has the first 16 spines (1S to 16S). The second set of uplinks connects to spines S17 to S32, as shown. Using this approach, the network can scale to thousands of leaf servers. Figure 17 An implementation of a network with 256 leaf switches is illustrated.

[0073]

[0074] Table III, Parameters for a 3:1 oversubscribed two-tier FCN with 16 spine switches. For the number of switches and servers, the number of modules 400 required and their size in racks are shown.

[0075]

[0076] Table IV, Parameters of a two-layer FCN with 1:1 oversubscription and 32 spine switches

[0077] For the number of switches and servers, the number of required modules 400 and their sizes in rack units are shown.

[0078] As shown in Tables III and IV, using a two-tier network, this two-tier network can scale to support thousands of leaf switches that can interconnect tens or thousands of servers. Scaling beyond this number requires using a three-tier FCN. Figure 1 (b) shows the topology of a three-layer network with 16 spine switches and 32 leaf switches. In a three-layer network, the spine does not need to be connected to all leaf switches, but to a group of leaf switches, called PODs.

[0079] Module 400 can also be used to implement a three-layer FCN, such as Figure 18 、 19 and 20. Figure 18In FIG, a three-layer network with 256 spine switches and 512 leaf switches is shown. Each POD 900 of the network has 16 switching meshes and leaf switches. The mesh of each POD can be fully implemented with four stacking modules 400, as previously shown (see FIG. Figure 10 (d)). Since there are 32 PODs 900, 128 modules 400 are required to implement the first part of the network (the fabric switch to the leaf switch).

[0080] exist Figure 18 In the embodiment, the second layer spine-to-leaf switch is implemented using the module 400. Since each spine switch only needs to be connected to one leaf switch in the POD, 32x256=8192 ports can be arranged in 512 modules 400, as shown in FIG. Figure 19 The leaf and spine switches are interconnected as shown in Figure 20 and Figure 21 As shown. Obviously, due to the limitation of rack space, the module stack needs to be installed in multiple racks. Following the above method, the uplinks of the leaf switches in each POD are horizontally filled with modules. For example, the four MPO uplinks from the first leaf switch of POD 1, namely L1p1, L1p1, L1p1 and L1p1, occupy the first MPO port of the first module 400. The four MPO uplinks from the leaf switch 16 of POD 32, namely L16p32, L16p32, L16p32 and L16p32, occupy the last four MPO ports of the last module 400 in the stack. From the opposite side of the stack, the columns of the module stack are connected to the line card MPO ports of the spine switch. For example, as Figure 20 As shown, line card MPO port S1c1 of spine switch 1 connects to the eight MPO ports in the first column of the stack. MPO port S1c2 of the second line card of the same switch connects to the eight MPO ports in the second column of the stack. MPO port S16c16 of line card 16 of the last spine switch connects to the eight ports in the last column of the stack.

[0081] This three-layer switching network has 256 spines (or 16 racks with 16 line cards) and 512 leaf switches, requiring 256 modules 400, with the equivalent rack space of the modules 400 being equal to or less than 90RU. Tables V and VI show how to scale this network with 3:1 and 1:1 oversubscription, as well as the required number of modules 400 and rack space.

[0082] In general, the module 400 and the disclosed interconnection method for two-tier and three-tier FCNs simplify the deployment of optical networks of different sizes and configurations. Because the cable groups representing the uplinks / downlinks of the same switch are closely connected, and also because of the high degree of meshing in the network, the risk of interconnection errors during deployment is greatly reduced. For example, in Figure 19 In , all Lipj connections are interchangeable, where i is the leaf uplink index ranging from 1 to 4 and j is the POD index ranging from 1 to 32. During network deployment, unexpected changes from L1p1 to L1p2, L1p1 to L1p3, L1p1 to L1p4, or any combination within the group will not have an impact on network operation. The topology will still use the same number of paths and the same allocated bandwidth to connect all leaf switches from the POD to the spine switch. Similarly, in Figure 20 All ridges in any row are interchangeable, as can be seen from Figure 9 The level of redundancy provided by the stacking of modules 400 greatly reduces the risk of switch fabric failure or performance degradation caused by errors in the interconnect.

[0083]

[0084] Table V Parameters for a Layer 3 FCN with 1:3 oversubscription of 256 spine switches (16 chassis with 16 line cards) For the number of switches and servers, the number of modules 400 required and their size in racks are shown.

[0085]

[0086] Table VI Parameters for two-tier FCNs with 1:1 oversubscription for 1024 spine switches (64 racks with 16 line cards) The number of modules 400 required and their size in racks are shown for the number of switches and servers.

[0087] Although the present invention has been described as having a preferred design, the present invention may be further modified within the spirit and scope of the present disclosure. Therefore, this application is intended to cover any variations, uses, or modifications of the present invention using its general principles. Further, this application is intended to cover departures from the present disclosure that come within known or customary practice in the art to which the present invention pertains and that fall within the limits of the appended claims.

Claims

1. A switching network module, characterized in that: The switching network module has multiple multi-fiber connector interfaces, some of which can be connected to network devices in the network using multi-fiber cables, the network including an internal grid implemented in two levels, wherein the first level is configured to rearrange the individual optical fibers in different optical fiber groups, and the second level is configured to recombine the individual optical fibers in the different optical fiber groups, further wherein the optical path of each transmitter and receiver is matched to provide appropriate optical connection from the transmitting optical fiber to the receiving optical fiber, and wherein complex arbitrary network topologies can be achieved by reducing point-to-point interconnections by at least 1 / N, where N = the number of channels per multi-fiber connector interface.

2. The switching network module according to claim 1, characterized in that: The switch fabric network modules are further configured to be stacked to provide two-level or three-level CLOS network topologies of various switch bases.

3. The switching network module according to claim 1, characterized in that: The switch network module is further configured to implement networks with different levels of oversubscription ranging from 1:1 to 1:

12.

4. The switching network module according to claim 1, characterized in that: The switch fabric network module is further configured to scale the optical network from eight switches to thousands of leaf switches.

5. The switching network module according to claim 1, characterized in that: The switch network module is further configured to provide redundant paths, thereby reducing the risk of switch network failure due to interconnection errors.

6. The switching network module according to claim 1, characterized in that: The switch fabric network module is further configured to have a small form factor, enabling three modules to be stacked in one RU, thereby allowing up to 132 modules to be stacked per rack.

7. The switching network module according to claim 1, characterized in that: The switch network module further includes a machine-readable tag that, when read by a tag reader such as a laser scanner or camera, is capable of providing an interconnection map of the network to a portable device.

8. The switching network module according to claim 2, characterized in that: The switching network module is further configured to effectively distribute the traffic load of the switch.

9. The switching network module according to claim 2, characterized in that: The interconnect ports use multi-fiber connectors with 4 to 32 fibers.

10. The switching network module according to claim 2, characterized in that: The interconnect ports use multi-fiber connectors of different form factors such as CS, SN, MPO, SN-MT, MMC.

11. The switching network module according to claim 2, characterized in that: Each fiber interconnection can transmit signals of different wavelengths in both co-transmission and reverse transmission bidirectionally.

12. A switching network, characterized in that: The switching network includes a stack of switching network modules, each module having multiple multi-fiber connector interfaces, and further each module is combined with an internal grid, which is implemented in two or more levels for rearrangement of fiber groups, wherein complex arbitrary network topologies can be achieved by reducing point-to-point interconnections by at least 1 / N, where N = the number of channels per multi-fiber connector interface.

13. The switching network according to claim 12, characterized in that The switching network is further configured to scale the optical network from eight switches to thousands of leaf switches or beyond.

14. The switching network according to claim 12, wherein: The switching network is configured to provide redundant paths, thereby reducing the risk of switching network failure due to interconnection errors.

Images