Physical layer interface with redundant data paths
By introducing multiplexers and integrated circuit modules into the physical layer interface equipment of the data center, intelligent switching and error correction of data paths are achieved, the problem of rising failure rates in the data center is solved, and the reliability and efficiency of services are improved , reducing the cost of redundant solutions.
Patent Information
- Application Number
- CN202011611833.3
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Priority Date
- 2020-06-17
- Filing Date
- 2020-12-30
- Publication Date
- 2024-11-15
- Estimated Expiration
- 2040-12-30
AI Technical Summary
In cloud computing data centers, the failure rate of hardware and software components continues to rise, resulting in an increase in the risk of service interruption. The existing redundant solutions are costly and inefficient, making it difficult to effectively deal with failures in hyperscale data centers.
Design a physical layer interface device, including transmitters and receivers for primary and secondary data paths, and multiplexers, to switch data paths when detecting failures through the controller, ensure service continuity, and through integration The circuit module performs error correction and data protection, realizing re-time and selective transmission of data streams.
It realizes rapid switching of data paths when a failure occurs, reduces service interruption time, improves the reliability and efficiency of the data center, reduces the cost of redundant solutions, and can complete state transitions in a few nanoseconds, ensuring the stability of data communication. .
Smart Images

Figure CN113810317B_ABST
Abstract
Description
Background Art
[0001] Data centers used for cloud computing must run customer applications without interruption. However, both hardware and software components inevitably fail at a rate characterized by their mean time to failure. As data center infrastructure becomes more complex, the total failure rate rises rapidly, and for hyperscale data centers, the number of failures becomes unmanageable.
[0002] One way to address this problem is to provide some form of redundancy that enables operations to continue even while a fault is identified and repaired. But when expressed in hardware, redundancy can take the form of active components and passive backup components that are ready to take over if an active component were to fail, thereby preventing service interruption.
[0003] While such redundancy is beneficial, it would be very expensive or inefficient to simply provide a backup for every hardware component. Instead, it is desirable to provide redundancy only where it is most beneficial to do so.
[0004] Where existing components can be configured to provide such redundancy without duplicating entire components, greater efficiencies can be achieved. For example, if a network switch is equipped with sufficient internal redundancy, using such a switch to achieve the desired performance level will be more efficient than providing a passive backup switch for each active switch and the additional connecting cables required by such an arrangement. Summary of the invention
[0005] Therefore, an embodiment of a physical layer interface device is disclosed herein, comprising: a first transmitter and a first receiver for a primary data path; a second transmitter and a second receiver for a secondary data path; a third transmitter and a third receiver for a non-redundant data path; and a multiplexer. The third receiver is coupled to simultaneously provide a data stream received from the non-redundant data path to the first and second transmitters, and the multiplexer provides the third transmitter with a selected one of the data stream received via the primary data path and the data stream received via the secondary data path.
[0006] A first disclosed network switch embodiment includes: a plurality of physical layer interfaces, an internal switch structure, and at least one controller. The plurality of physical layer interfaces each transmit a data stream packet from a corresponding network port to a corresponding primary data path and a secondary data path, and transmit the data stream packet from a selected one of the corresponding primary data path and the secondary data path to the corresponding network port. The internal switch structure guides the data stream packet between the plurality of physical layer interfaces. The controller configures at least one of the plurality of physical layer interfaces to convert from the primary data path corresponding to the physical layer interface to the secondary data path corresponding to the physical layer interface when the controller detects a fault associated with the corresponding primary data path.
[0007] A second disclosed network switch embodiment also includes: a plurality of physical layer interfaces, an internal switch structure, and at least one controller. In this embodiment, the plurality of physical layer interfaces each transmit data flow packets from a corresponding non-redundant data path to a corresponding primary network port and a secondary network port, and transmit data flow packets from a selected one of the corresponding primary network port and the secondary network port to the corresponding non-redundant data path. An internal switch structure guides data flow packets between the plurality of physical layer interfaces. The controller configures at least one of the plurality of physical layer interfaces to convert from the primary network port corresponding to the physical layer interface to the secondary network port corresponding to the physical layer interface when the controller detects a fault associated with the corresponding primary network port.
[0008] Each of the foregoing embodiments may be implemented individually or in combination, and may be implemented with any one or more of the following features in any suitable combination: 1. A controller that determines a link condition of a primary data path and controls a multiplexer to select the primary data path or a secondary data path based on the link condition. 2. Each of the first, second, and third transmitters transmits a retimed data stream from a corresponding one of the first, second, and third receivers without error correction and regenerated error correction code protection. 3. The physical layer interface includes one or more integrated circuit modules that at least correct errors, perform packet integrity checks, and regenerate error correction code protection on a data stream provided from the third receiver to the first transmitter. 4. The second transmitter transmits at least a data stream from the third receiver without error correction and regenerated error correction code protection. 5. The physical layer interface device includes: a second integrated circuit module between the multiplexer and the third transmitter of the one or more integrated circuit modules, the second integrated circuit module generating error correction code protection for a selected one of the data streams. 6. The physical layer interface device includes: a third integrated circuit module of the one or more integrated circuit modules coupled to the second receiver, the third integrated circuit module performing error correction on the data stream received via the secondary data path. 7. The controller is one of a plurality of controllers, and each of the plurality of physical layer interfaces includes a corresponding one of the plurality of controllers to determine the state of the corresponding network port and select between the corresponding primary data path and the secondary data path based on the state. 8. The controller detects a fault by comparing an error rate or a packet loss rate with a predetermined threshold. 9. The switch structure replicates data flow packets to deliver the data flow packets to each of the plurality of physical layer interfaces via both the primary data path and the secondary data path. 10. The controller adjusts the routing plan of a data packet directed to at least one of the plurality of physical layer interfaces to direct the data packet via the secondary data path instead of the primary data path. BRIEF DESCRIPTION OF THE DRAWINGS
[0009] FIG. 1 is a block diagram of a conventional network switch.
[0010] Figure 2 is a block diagram of a network switch with internal redundancy added.
[0011] Figure 3 is a block diagram of a network switch that adds connection redundancy.
[0012] Figure 4 is a block diagram of a first illustrative physical layer interface.
[0013] Figure 5 is a block diagram of an illustrative multi-channel physical layer interface.
[0014] Figure 6is a flow chart of an illustrative reliability enhancement method.
[0015] Figure 7 is a block diagram of a second illustrative physical layer interface.
[0016] Figure 8 is a block diagram of a third illustrative physical layer interface. DETAILED DESCRIPTION
[0017] Although specific embodiments are presented in the drawings and the following description, it should be kept in mind that they do not limit the present disclosure. Instead, they provide a basis for one of ordinary skill in the art to identify alternatives, equivalents, and modifications that are included within the scope of the appended claims.
[0018] FIG. 1 is a block diagram of a conventional network switch 100 having a port AN to which a network cable can be connected to transmit data streams to or from other network nodes (e.g., a server or other switch). Physical layer interfaces 102A-102N couple each port to an internal switch fabric 104 to provide at least channel signal transmission and reception hardware, but potentially also provide channel synchronization, error correction encoding / decoding, data scrambling / descrambling, multiplexing / demultiplexing, and data framing / extraction. The internal switch fabric 104 receives incoming data packets from the physical layer interface 102, checks the header information contained therein to determine at least the destination address information, and directs the data packets to the appropriate physical layer interface as outgoing data packets. Publicly available literature discloses many suitable implementations for the internal switch fabric, ranging from cross-bar switches to bus-based architectures. See, for example, US6259699 "System architecture for and method of processing packets and / or cells in a common switch" and US7079525 "Network switch having a hybrid switch architecture".
[0019] The internal switch fabric 104 typically includes one or more controllers that coordinate the operation of the other components to provide the desired functionality, and may also monitor the performance of the physical layer interfaces 102 and adjust their operating parameters or configurations. The controllers may be programmable general purpose processors configured by firmware, or they may take the form of application specific integrated circuits (ASICs). Typically, most of the components of the internal switch fabric 104 need to handle high data transfer volumes with minimal latency, and are accordingly implemented as ASICs.
[0020] When a switch is in service in a data center, it operates nonstop at relentless speeds to direct staggering amounts of data traffic. No matter how reliable the switch components are, failures should be expected. Such failures are often localized, initially impairing the operation of only a small subset of ports (e.g., 1 or 2 ports). However, those 1 or 2 ports may be the only (multiple) connections a server has to the network.
[0021] In order to enhance the reliability of each switch port, this paper proposes a switch 200 using an enhanced physical layer interface design and method. Figure 2 As shown, the internal switch fabric 104 is connected to each of the physical layer interfaces 202(A)-202(N / 2) via the following redundant paths: a primary path and a secondary path. In the initial (default) state, only the primary path is selected for transmitting data packets between the physical layer interface and the internal switch fabric. If performance on a given primary path becomes degraded, the associated physical layer interface 202 can switch (or be switched) to a standby state, in which only the secondary path is selected for transmitting data packets between the physical layer interface and the internal switch fabric. In either case, data communications via the external switch port can continue through the non-redundant side of the physical layer interface. The use of the standby state can provide time to diagnose and correct performance degradation problems.
[0022] The primary path and the secondary path can be coupled to relatively independent portions of the internal switch fabric 104, making it unlikely that both paths will be affected by a single or localized component failure. In some contemplated embodiments, when a failure is detected, the internal switch fabric 104 simply reroutes the data stream from the failed primary data path to the associated secondary data path. A failure can be detected as a bit error rate, a symbol error rate, or a packet loss rate exceeding a predetermined threshold. In other contemplated embodiments, the internal switch fabric 104 replicates data packets addressed for each given port to provide them on both the primary data path and the secondary data path, so that the switch from the primary data path to the secondary data path can be accomplished using only reconfiguration of the corresponding physical layer interface.
[0023] Note that if the internal switch fabric 104 only supports the same number of paths as in FIG. 1 , then Figure 2 The number of switch ports in the embodiment will be halved in exchange for increased reliability. Conversely, if the internal switch structure is scaled to support twice the paths of Figure 1, then Figure 2 An embodiment maintains the number of switch ports.
[0024] In relevant cases where internal switch components are proven to have a lower failure rate (longer mean time to failure) than port connectors and associated network cables, Figure 3 The redundant and non-redundant sides of the physical layer interface are reversed as shown. The illustrative network switch 300 shows each of the physical layer interfaces 302A-302N coupled to the following redundant external interfaces: a primary port AN and a secondary port A'-N'. (It is expected that the redundant ports will be connected to the same external node via redundant cables.) The internal switch fabric 104 is coupled to each of the physical layer interfaces via a single non-redundant path. As previously described, in the initial (default) state, only the primary port is selected for transmitting data streams to and from an external node (e.g., a server or another switch). If the performance on a given primary port becomes degraded, the physical layer interface 202 can switch (or be switched) to a standby state, in which the associated secondary port is selected for transmitting data streams to and from the associated external node. In both states, data packets will continue to be transmitted to and from the internal switch fabric 104 via the non-redundant path.
[0025] Figure 4 An illustrative implementation of a physical layer interface is shown that couples a non-redundant path ("path A0") to two redundant paths ("path P0" and "path S0"). The discussion here goes into unnecessary detail, but additional implementation details regarding the physical layer interface components can be found, for example, in commonly owned U.S. Patent 10,212,260, entitled "SerDes Architecture with a Hidden Backchannel Protocol" ("Sun"), which is hereby incorporated by reference in its entirety. Each of the data paths discussed herein can be a multi-lane data path operating in accordance with the Ethernet standard (IEEE Std 802.3-2015 or later), in which case any switching of the data paths will be applied simultaneously on all lanes.
[0026] Receiver 402 and transmitter 404 are connected to a non-redundant data path (shown here as path A0 of an external data port) to receive and send data streams. According to the Ethernet standard, and as described more fully in Sun's patent, receiver 402 and transmitter 404 can implement the following sublayers of the Ethernet protocol: physical medium attachment (PMA), and physical medium dependence (PMD). For the receiver, this will include equalization, symbol detection, serial to parallel conversion, and channel de-skewing. For the transmitter, this will include parallel to serial conversion, symbol modulation, pre-equalization, and transmission.
[0027] The MAC module 406 implements the forward error correction layer (FEC), physical coding sublayer (PCS), and medium access control (MAC) sublayers of the Ethernet protocol, providing, among other things, symbol detection / decoding, error correction (for incoming data) and regeneration of error correction code protection (for outgoing data), as well as packet integrity verification (for incoming data) and checksum generation (for outgoing data).
[0028] A first-in-first-out (FIFO) buffer 408 couples the MAC module 406 to a second MAC module 410 for the primary data path P0. A second receiver 412 and a transmitter 414 couple the second data link module 410 to the primary data path P0 to implement PMA, PMD functions.
[0029] For the secondary data path S0, the third receiver 422 provides parallel detection of the symbol data stream to the third data link module 426, and the second set of FIFO buffers 428 couples the third data link module 426 to the first data link module 406 via a multiplexer 430. When the primary data path is active, the multiplexer 430 selects the FIFO buffer 408, and when the secondary data path is active, the multiplexer 430 selects the FIFO buffer 428. In either state, the FIFO buffer 428 can couple the parallel detected symbol data stream from the receiver 402 to the third transmitter 424 for transmission on the secondary data path S0.
[0030] In the default state where all data paths are available, the multiplexer selects the primary data path. The non-redundant data path is provided with a data stream received via the primary data path P0, and the data stream received via the secondary data path S0 (if any) is discarded. The multiplexer state can be controlled by an internal register of the physical layer interface device, and if an error is detected internally, the multiplexer state can be set by the physical layer device, or the multiplexer state can be set by an external controller (e.g., a switch fabric controller). The physical layer interface device can determine the link status or network port status based on, for example, a bit error rate, a symbol error rate, a packet loss rate, etc., thereby determining a valid status or state when the error rate is below a predetermined threshold. For example, the MAC module 410 can detect symbol errors and packet errors to monitor the error rate, and thereby determine whether the fault is associated with the primary data path (or alternatively with the currently selected data path).
[0031] Data streams received via the non-redundant data path A0 are broadcasted through both the primary port P0 and the secondary port S0. Data streams arriving at the primary port P0 pass through data link modules 406, 410, which provide FEC removal and regeneration and packet integrity checking. Data streams arriving at the secondary port follow a "retiming" path, which lacks FEC / PCS / MAC functionality. This path relies on external nodes and the internal switch fabric for error correction and packet integrity checking for the links between them; in this state, the physical layer interface does not provide such protection for the individual link segments.
[0032] We note here that when the primary data path is active (selected), the secondary data path may experience multiple interruptions without affecting the traffic between the non-redundant path and the primary data path. If the primary path P0 fails due to a hardware or software failure, the data stream received via the non-redundant data path A0 is still broadcast to the secondary data path S0, and any data received via the secondary data path is transmitted to the multiplexer 430, which can select the data for transmission via the non-redundant data path. The physical layer interface or external controller can detect the failure and change the state of the multiplexer 430. The transition between states is fast, i.e., on the order of a few nanoseconds. During the transition, the link conditions on the non-redundant data path and the secondary data path remain stable.
[0033] Although link conditions can generally tolerate one or two packets being truncated which may result from an asynchronous switching of the multiplexer 430, the physical layer interface can be easily arranged for synchronous switching. The physical layer interface can monitor packet header information so that switching can begin after the end of the packet from the primary path and complete when the packet from the secondary path begins. An idle mode can be used to maintain the link during the switching interval.
[0034] The transition can be associated with an error code or alarm signal in an internal register of the physical layer interface, causing the switch controller to transmit an alarm message to service personnel. Because the secondary data path is operational, service personnel have time to diagnose and resolve the cause of the primary data path failure while the server remains connected and available.
[0035] When the primary data path resumes operation, the MAC modules 406, 410 can resume or continue to transmit the data stream received via the non-redundant data path A0 to the primary data path P0. The data stream received via the primary data path P0 (if any) is transmitted to the multiplexer 430, but because the secondary path is selected, the data stream is discarded. However, at this time, the physical layer interface is ready to switch back to the primary data path when the instruction comes from the controller. As previously described, the state transition is very fast, that is, on the order of a few nanoseconds.
[0036] although Figure 4 The example of provides asymmetric FEC / PCS / MAC functionality for data traveling to and from the secondary path S0, but alternative embodiments that provide symmetric processing are contemplated. Figure 7 An alternative embodiment is shown that includes the previously discussed MAC modules 406, 410 to provide FEC / PCS / MAC functionality between the primary data path and the non-redundant data path, but further includes MAC modules 706, 710 to provide the same functionality between the secondary data path and the non-redundant data path. Using this arrangement of MAC modules, each of the data paths can operate with a certain degree of independence. Note that the multiplexer 430 ( Figure 4 ), and instead a multiplexer 730 is provided at the input of the transmitter 404 to select between the data stream from the MAC module 406 and the data stream from the MAC module 706.
[0037] Figure 8 Another alternative embodiment is shown that omits the FEC / PCS / MAC functionality altogether, uses the receiver to perform equalization and symbol detection, and uses the transmitter to resend detected symbols, thereby providing a retiming function without error correction or packet integrity verification. A multiplexer 830 is provided at the input of the transmitter 404 to select between data streams received via the primary path and the secondary path. Such a retiming implementation may employ asynchronous switching between the primary and secondary data paths, potentially resulting in packet truncation and associated link condition transients.
[0038] Although the aforementioned digital embodiments are preferred, it is also feasible to implement the physical layer interface as an analog signal repeater and a multiplexer that selects between the primary and secondary data paths, providing filtering and signal amplification without symbol detection and remodulation. A physical layer interface that supports multi-interface operation (analog repeater, digital retimer, MAC function, and asymmetric combinations thereof) can provide additional flexibility to enable mitigation of degradation in the event of hardware failures.
[0039] To support multi-port or non-redundant data paths, multiple instances of the physical layer interface may be included in a single packaged integrated circuit as a physical layer interface device. Figure 5 An example of four instances of a physical layer interface included in a single device is shown. It is contemplated that each instance operates independently of the other instances, such that, for example, three of the instances may continue to operate in a default state while a fourth instance begins operating using a selected secondary data path.
[0040] Figure 6 is a flow chart of an illustrative reliability enhancement method that can be implemented by a controller in a switch having a disclosed physical layer interface. (The controller can be part of an internal switch fabric or included in one or more of the physical layer interfaces) In block 602, the physical layer interface defaults to an active state, in which data received via a non-redundant path is copied to both a primary data path and a secondary data path, and data sent via the non-redundant data path is received via the primary data path.
[0041] In block 604, the active state is retained until a failure is detected on the primary data path. In block 606, the physical layer interface is transitioned to a standby state in which data received via the non-redundant path is copied to both the primary data path and the secondary data path, and data sent via the non-redundant data path is received via the secondary data path. An alarm is preferably communicated to initiate fault correction.
[0042] In block 606, the standby state is retained until the failure is corrected, at which point the method returns to block 602. It is expected that the state transition will be fast so that the stability of each data path is preserved.
[0043] It is expected that the foregoing embodiments facilitate the practical and economical implementation of path redundancy. Once the above disclosure is fully understood, numerous alternatives, equivalents, and modifications will become apparent to those skilled in the art. For example, the embodiments described above provide redundancy in the form of a single secondary data path for each primary data path, but those of ordinary skill in the art will recognize that the disclosed principles can be easily expanded to provide multiple secondary data paths to further increase the redundancy of each primary data path. It is intended that the claims be interpreted as covering all such alternatives, equivalents, and modifications that are included within the scope of the appended claims.
Claims
1. A physical layer interface device, include: a first transmitter and a first receiver for a primary data path; a second transmitter and a second receiver for a secondary data path; a third transmitter and a third receiver for a non-redundant data path, the third receiver coupled to simultaneously provide the first and second transmitters with a data stream received from the non-redundant data path; as well as a multiplexer that provides the third transmitter with a selected one of a data stream received via the primary data path and a data stream received via the secondary data path; as well as one or more integrated circuit modules to correct errors and regenerate error correction code protection for a data stream provided from the third receiver to the first transmitter, Wherein the second transmitter transmits the data stream from the third receiver without error correction and without regenerated error correction code protection.
2. The physical layer interface device as claimed in claim 1, further comprising: include: A second integrated circuit module of the one or more integrated circuit modules between the multiplexer and a third transmitter, the second integrated circuit module generating error correction code protection for the selected one of the data streams.
3. The physical layer interface device as claimed in claim 2, further comprising: include: A third integrated circuit module of the one or more integrated circuit modules is coupled to the second receiver, the third integrated circuit module performing error correction on a data stream received via the secondary data path.
4. A network switch, include: a plurality of physical layer interfaces, each physical layer interface transmitting a data stream packet from a corresponding network port to a corresponding primary data path and a secondary data path, and transmitting a data stream packet from a selected one of the corresponding primary data path and the secondary data path to the corresponding network port; an internal switch fabric that directs data flow packets between the plurality of physical layer interfaces; as well as a controller that configures at least one of the plurality of physical layer interfaces to switch from the primary data path corresponding to the physical layer interface to a secondary data path corresponding to the physical layer interface when the controller detects a fault associated with the corresponding primary data path, Each of the plurality of physical layer interfaces comprises: a first transmitter and a first receiver for their respective primary data paths; a second transmitter and a second receiver for their respective secondary data paths; a third transmitter and a third receiver for their respective network ports, the third receiver being coupled to simultaneously provide the first and second transmitters with data streams received from the network ports; a multiplexer that provides the third transmitter with a selected one of a data stream received via the primary data path and a data stream received via the secondary data path; and One or more integrated circuit modules are configured to at least perform error correction, check packet integrity, and regenerate error correction code protection for a data stream provided from the third receiver to the first transmitter.
5. The network switch according to claim 4, It is characterized in that The controller detects the failure by comparing the packet loss rate with a predetermined threshold.
6. The network switch according to claim 4, It is characterized in that The controller configures the internal switch fabric to replicate a data flow packet to deliver the data flow packet to each of the plurality of physical layer interfaces via both the primary data path and the secondary data path.
7. The network switch according to claim 4, It is characterized in that The controller causes the internal switch fabric to adjust a routing plan of a data packet directed to at least one of the plurality of physical layer interfaces, the adjusted routing plan directing the data packet via the secondary data path instead of the primary data path.
8. The network switch according to claim 4, It is characterized in that The controller is one of a plurality of controllers, and each of the plurality of physical layer interfaces includes a corresponding one of the plurality of controllers to determine a status of a corresponding network port and to select between a corresponding primary data path and a secondary data path based on the status.
9. The network switch according to claim 4, It is characterized in that In each of the plurality of physical layer interfaces, the second transmitter sends a data stream from the third receiver without error correction and regenerated error correction code protection.
10. The network switch according to claim 9, It is characterized in that Each of the plurality of physical layer interfaces further comprises: a second integrated circuit module of the one or more integrated circuit modules between the multiplexer and a third transmitter, the second integrated circuit module generating error correction code protection for a selected one of the data streams.
11. The network switch according to claim 10, It is characterized in that Each of the plurality of physical layer interfaces further comprises: a third integrated circuit module of the one or more integrated circuit modules coupled to the second receiver, the third integrated circuit module performing error correction on a data stream received via the secondary data path.
12. A network switch, include: a plurality of physical layer interfaces, each physical layer interface transmitting a data stream packet from a corresponding non-redundant data path to a corresponding primary network port and a secondary network port, and transmitting a data stream packet from a selected one of the corresponding primary network port and the secondary network port to the corresponding non-redundant data path; an internal switch fabric that directs data flow packets between the plurality of physical layer interfaces; as well as a controller, the controller configuring at least one of the plurality of physical layer interfaces to switch from the primary network port to which the physical layer interface corresponds to to a secondary network port to which the physical layer interface corresponds when the controller detects a fault associated with the corresponding primary network port, Each of the plurality of physical layer interfaces comprises: a first transmitter and a first receiver for their respective primary data paths; a second transmitter and a second receiver for their respective secondary data paths; a third transmitter and a third receiver for their respective network ports, the third receiver being coupled to simultaneously provide the first and second transmitters with data streams received from the network ports; a multiplexer that provides the third transmitter with a selected one of a data stream received via the primary data path and a data stream received via the secondary data path; and One or more integrated circuit modules are configured to at least perform error correction, check packet integrity, and regenerate error correction code protection for a data stream provided from the third receiver to the first transmitter.
13. The network switch according to claim 12, It is characterized in that The controller detects the fault by comparing the error rate with a predetermined threshold.
14. The network switch according to claim 12, It is characterized in that The controller is one of a plurality of controllers, and each of the plurality of physical layer interfaces includes a corresponding one of the plurality of controllers to determine a status of a corresponding primary network port and to select between the corresponding primary network port and a secondary network port based on the status.
15. The network switch according to claim 12, It is characterized in that In each of the plurality of physical layer interfaces, the second transmitter sends a data stream from the third receiver without error correction and without regenerated error correction code protection.
16. The network switch according to claim 15, It is characterized in that Each of the plurality of physical layer interfaces further comprises: a second integrated circuit module of the one or more integrated circuit modules between the multiplexer and a third transmitter, the second integrated circuit module generating error correction code protection for a selected one of the data streams.
17. The network switch according to claim 16, It is characterized in that Each of the plurality of physical layer interfaces further includes a third integrated circuit module of the one or more integrated circuit modules coupled to the second receiver, the third integrated circuit module performing error correction on a data stream received via the secondary data path.
