Multi-host networking system and method

Abstract

Systems and methods are provided. An illustrative system includes a first computing node having a first processing unit, a first computing node port, and a first peripheral component interconnect bus configured to carry data between the first processing unit and the first computing node port. The system can also include a multi-host network interface controller having a first multi-host port, where the first multi-host port is configured to connect with the first computing node port via a first peripheral component networking cable, a network port, where the network port is configured to receive a network interface of a networking cable, and a processing circuit configured to translate and carry data between the first multi-host port and the network port.

Description

TECHNICAL FIELD

[0001] The present disclosure relates generally to networking, and in particular to a multi-host networking solution. BACKGROUND

[0002] Many currently available servers are built in configurations where multiple compute node elements are confined in a common space and share certain resources (e.g., power, space, and thermal control resources) among the nodes. Although the compute nodes share power, space, and / or thermal control resources, the compute nodes typically each independently communicate with a top-of-rack (TOR) switch. As the compute nodes run at higher speeds and utilize more power, the compute nodes will generate more heat, making the overall density of the enclosure chassis difficult to manage. It is especially difficult to integrate more compute nodes in the common space without negatively impacting other compute nodes already in the common space. SUMMARY

[0003] To address the challenges described above, system designers are struggling to integrate high performance network interface controller (NIC) cards in the conventional way of standard half-height half-length peripheral component interconnect express (PCIe) cards to fill the available slots. The introduction of additional NIC cards is particularly difficult because such introduction increases the speed of the system fans to the maximum allowed to enable sufficient cooling conditions. Unfortunately, the noise caused by the fans when running at maximum speed typically interferes with the signal transmission within the server enclosure.

[0004] Many modern servers are built in a constellation enclosure that includes a 2U chassis of four network hosts (e.g., compute nodes). These servers and similar implementations can be very dense, requiring advanced thermal solutions that can be very complex and costly if fan noise is expected to remain below a certain threshold.

[0005] In many existing solutions, a compute node communicates with its environment using a standard PCIe NIC connected to a TOR switch, connecting the rack to the rest of the network. This connection between the compute node and the TOR switch utilizes the PCIe communication standard / protocol and requires the TOR switch to be equipped with a dedicated port for each of the rack compute nodes. Since the compute node is a standalone entity and needs to be served as a standalone entity, it contains its own NIC card and connects directly to the TOR.

[0006] In some cases, storage modules can aggregate several compute nodes into a single unit, enabling the sharing of certain resources between nodes. This resource sharing primarily occurs in the areas of power, space, and thermal resources. Network resources are not typically shared between compute nodes. Embodiments of the present disclosure propose a system in a method where network resources are shared between multiple compute nodes while enabling full maintainability of compute nodes as independent entities.

[0007] More specifically, embodiments of the present disclosure propose a multi-host NIC that will enable the sharing of network resources between multiple compute nodes while still maintaining compute nodes as independent entities within a housing unit, enabling each compute node to be independently maintained.

[0008] A multi-host NIC card is proposed that, in addition to network ports, includes several external multi-host port connectors. In some embodiments, each of the multi-host port connectors (also referred to herein as multi-host ports) can enable connection and communication with other compute nodes residing in the same housing as the multi-host NIC. One, some, or all of the compute nodes can also include a supplemental card having a re-timer unit (e.g., a PCIe re-timer unit) that enables routing of PCIe buses from the compute nodes to external peripheral component interconnect cables (e.g., external PCIe bus connectors).

[0009] In some embodiments, each of the compute nodes can be provided with a supplemental card, and then each of the compute nodes can reside in a common server housing having the proposed multi-host NIC. The multi-host NIC can be configured as a networking sled or blade to accommodate the structure of the server housing. Peripheral component interconnect cables can then attach each of the supplemental cards to the multi-host NIC, enabling connection and communication between the compute nodes and the multi-host NIC.

[0010] The multi-host NIC can then be connected to the TOR switch of the server rack. In some embodiments, all communication between the TOR switch and each of the compute nodes flows through the multi-host NIC. More specifically, all communication between the compute nodes and the broader communication network can flow through a single port of the TOR switch.

[0011] An example server enclosure can include four (4) compute nodes and a network tray or network sled that includes a proposed multi-host NIC. As described above, each of the compute nodes can incorporate an auxiliary card instead of a dedicated NIC card as in previous server configurations. The auxiliary card can connect to a multi-host NIC that connects directly to only one TOR switch port. In this configuration, the compute nodes share the common network resources (e.g., multi-host NIC and / or TOR switch port) of the server rack. The proposed configuration is also useful for the overall performance of the compute nodes in the server enclosure because the compute nodes can synchronize (e.g., through operation of the multi-host NIC). For example, because the proposed multi-host NIC is used by multiple compute nodes, the compute nodes can operate at higher speeds and bandwidth, enabling the resources contained in the server enclosure to accommodate higher network bandwidth bursts. Moreover, because a single TOR switch port is used, dual-port NICs can be implemented, again enabling higher network bandwidth.

[0012] In an illustrative example, a system is disclosed that includes a first compute node that includes: a first processing unit; a first compute node port; and a first peripheral component interconnect bus configured to carry data between the first processing unit and the first compute node port. The disclosed system also includes a multi-host network interface controller that includes: a first multi-host port, wherein the first multi-host port is configured to connect with the first compute node port via a first peripheral component networking cable; a network port, wherein the network port is configured to receive a network interface of the networking cable; and processing circuitry configured to translate and carry data between the first multi-host port and the network port.

[0013] In another example, a server enclosure is disclosed that includes a first compute node that includes: a first auxiliary card; a first processing unit; a first compute node port; and a first peripheral component interconnect bus configured to carry data between the first processing unit and the first compute node port. The server enclosure is also disclosed to include a multi-host network interface controller that includes: a first multi-host port, wherein the first multi-host port is configured to connect with the first compute node port via a first peripheral component networking cable; a network port, wherein the network port is configured to receive a network interface of the networking cable; and processing circuitry configured to translate and carry data between the first multi-host port and the network port.

[0014] In yet another example, a method is disclosed, the method comprising: installing a multi-host network interface controller in a server enclosure, wherein the multi-host network interface controller comprises a first multi-host port, wherein the first multi-host port is configured to connect with a first compute node port of a first compute node via a first peripheral component networking cable, wherein the multi-host network interface controller further comprises a network port configured to receive a network interface of a networking cable, and wherein the multi-host network interface controller further comprises processing circuitry configured to translate data and carry data between the first multi-host port and the network port; installing a first daughter card in the server enclosure, wherein the first daughter card comprises the first compute node, and wherein the first daughter card comprises a first peripheral component networking cable bus configured to carry data between the first processing unit and the first compute node port; enabling a data flow between the first daughter card and the multi-host network interface controller via a first communication protocol; and enabling a data flow between the multi-host network interface controller and a communication network via a second communication protocol, thereby facilitating communication between the first daughter card and the communication network.

[0015] Additional features and advantages are described herein, and will be apparent from, the following DESCRIPTION and the figures. BRIEF DESCRIPTION OF DRAWINGS

[0016] The disclosure is described in connection with the appended drawings, which are not necessarily drawn to scale:

[0017] Figure 1A is a block diagram depicting an illustrative configuration of a system in accordance with at least some embodiments of the present disclosure;

[0018] Figure 1B is a block diagram depicting another illustrative configuration of a system in accordance with at least some embodiments of the present disclosure;

[0019] Figure 2 is a perspective view of a multi-host NIC connected with a plurality of compute nodes in accordance with at least some embodiments of the present disclosure; and

[0020] Figure 3 is a flowchart depicting a method of configuring a system to utilize a multi-host NIC in accordance with at least some embodiments of the present disclosure. DETAILED DESCRIPTION

[0021] The following description provides implementation(s) and is not intended to limit the scope, applicability or configuration of the claims in any way. Rather, the ensuing description will provide those skilled in the art with enabling descriptions of the implementation(s) and alternatives. Various changes can be made to the implementations described and the same can be carried out in a manner known to those skilled in the art.

[0022] As can be appreciated from the following description, and for computational efficiency reasons, the components of the system can be arranged in any suitable location within the network of distributed components without affecting the operation of the system.

[0023] Further, it should be appreciated that the various links of the connecting elements can be wired, trace, or wireless links, or any suitable combination thereof, or any other suitable known or later developed element(s) capable of supplying data to and / or communicating data from the connected elements. For example, the transmission medium used as a link can be any suitable electrical signal carrier, including coaxial cable, copper wire and optical fiber, electrical traces on a PCB, etc.

[0024] As used herein, the phrases "at least one", "one or more", "or", and "and / or", as used in any of the claims, are open-ended transitions. For example, each of the expressions "at least one of A, B, and C", "at least one of A, B, or C", "one or more of A, B, and C", "one or more of A, B, or C", "A, B, and / or C", and "A, B, or C" means A alone, B alone, C alone, A and B together, A and C together, B and C together, or A, B, and C together.

[0025] As used herein, the term "automatically" and variants thereof refer to any suitable process or operation that is completed without substantial human input when performing the process or operation. However, a process or operation can be automatic if the input is received prior to the performance of the process or operation, even if the performance of the process or operation uses substantial or non-substantial human input. Human input is considered substantial if it affects the manner in which the process or operation is performed. Human input that agrees to perform the process or operation is not considered "substantial".

[0026] As used herein, the terms "determine", "estimate", and "calculate" and variants thereof are used interchangeably and include any suitable type of methodology, process, operation, or technique.

[0027] Various aspects of the disclosure will be described with reference to the accompanying drawings, which are cited as illustrative of the idealized configurations.

[0028] Unless otherwise defined, all terms (including technical and scientific terms) used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this disclosure belongs. It will be further understood that terms, such as those defined in commonly used dictionaries, should be interpreted as having a meaning that is consistent with their meaning in the context of the relevant art and the present disclosure.

[0029] As used herein, the singular forms "a", "an" and "the" are intended to include the plural forms as well, unless the context clearly indicates otherwise. It will be further understood that the terms "comprises", "comprising", "includes" and / or "including", when used in this specification, specify the presence of stated features, integers, steps, operations, elements, and / or components, but do not preclude the presence or addition of one or more other features, integers, steps, operations, elements, components, and / or groups thereof.

[0030] Reference is now made to FIGS. 1 through Figure 3 Various systems and methods for operating a server enclosure having a multi-host NIC will be described in accordance with at least some embodiments of the present disclosure. The multi-host NIC can be configured to operate, translate, and / or route data flows between components of the server enclosure and other network resources of the server rack. For example, the multi-host NICs depicted and described herein can be configured to facilitate data packet transfer between a plurality of compute nodes and ports of a TOR switch. While certain embodiments will be described in connection with packet transmission or in connection with transmitting packets from one computing device to another using certain communication protocols, it will be appreciated that embodiments of the present disclosure are not so limited. The term "packet" as used herein should be interpreted to mean any suitable discrete quantity of digitized information, and the multi-host NICs described herein can be configured to operate on any such digitized information.

[0031] Reference is first made to Figure 1A and Figure 1B , an illustrative system 100 in accordance with at least some embodiments of the present disclosure is depicted. The system 100 can include a server rack 104 having one or more cards including a multi-host NIC 120 and one or more auxiliary cards 124a-d disposed therein. As Figure 1A shown in Figure 1BAs shown in FIG. 1, each card 120 can be disposed in a different server enclosure 112a-d. Illustratively, the first server enclosure 112a can include a multi-host NIC 120, the second server enclosure 112b can include a secondary card 160 belonging to a first compute node 124a, the third server enclosure 112c can include a secondary card 160 belonging to a second compute node 124b, the fourth server enclosure 112d can include a secondary card 160 belonging to a third compute node 124c, and the fifth server enclosure 112e can include a secondary card 160 belonging to a fourth compute node 124d. Thus, different compute nodes 124a-d and the multi-host NIC 120 can be provided as separate components, which can be held in the server rack 104 by separate server enclosures 112a-e.

[0032] The first server enclosure 112a can be configured to connect to the communication network 108 via one or more networking cables 180. In some implementations, the first server enclosure 112a can be connected to a connector plug 184 or port of the TOR switch 116 with a first networking cable 180. A second networking cable 180 can then connect the TOR switch 116 to the broader communication network 108. In some implementations, the networking cables 180 can correspond to the same type of networking cable used to connect the TOR switch 116 to the communication network 108. For example, the networking cables 180 can be fiber optic cables or any other physical means for carrying electrical and / or optical signals. It should be appreciated that the networking cables 180 used within the server rack 104 need not necessarily be the same type of networking cable as used to connect the TOR switch 116 to the communication network 108. For example, one of the networking cables 180 can carry optical signals, while another of the networking cables 180 can carry electrical signals. The type of networking cables 180 used outside of the first server enclosure 112a can depend on the nature of the communication network 108. For example, if the communication network 108 is a packet-based communication network (e.g., a communication network that uses the Internet Protocol (IP) or a similar packet-based communication protocol), the networking cable(s) 180 can be configured to support the communication protocol of the communication network 108. The first server enclosure 112a can be configured to connect to the communication network 108 via one or more networking cables 180. In some implementations, the first server enclosure 112a can be connected to a connector plug 184 or port of the TOR switch 116 with a first networking cable 180. A second networking cable 180 can then connect the TOR switch 116 to the broader communication network 108. In some implementations, the networking cables 180 can correspond to the same type of networking cable used to connect the TOR switch 116 to the communication network 108. For example, the networking cables 180 can be fiber optic cables or any other physical means for carrying electrical and / or optical signals. It should be appreciated that the networking cables 180 used within the server rack 104 need not necessarily be the same type of networking cable as used to connect the TOR switch 116 to the communication network 108. For example, one of the networking cables 180 can carry optical signals, while another of the networking cables 180 can carry electrical signals. The type of networking cables 180 used outside of the first server enclosure 112a can depend on the nature of the communication network 108. For example, if the communication network 108 is a packet-based communication network (e.g., a communication network that uses the Internet Protocol (IP) or a similar packet-based communication protocol), the networking cable(s) 180 can be configured to support the communication protocol of the communication network 108.

[0033] U.S. Patent No. 10,831,694 describes additional capabilities and details of suitable multi-host NICs 120 that can be used in accordance with at least some implementations of the present disclosure, the entirety of which is incorporated herein by reference.

[0034] The connector plug 184 of the TOR switch 116 can correspond to a physical, mechanical, electrical, and / or optical interconnect that enables the networking cable 180 to be physically inserted into and (electrically and / or optically) connected with components of the TOR switch 116. Illustratively, but not by way of limitation, the connector plug 184 can correspond to an Ethernet port, a fiber optic port, and / or the like.

[0035] The first server enclosure 112a (or the multi-host NIC 120 disposed therein) can also include a network port 132 that receives an opposite side of the networking cable 180. In some embodiments, the network port 132 can be responsible for carrying all communications / packets that are passed between the multi-host NIC 120 and the TOR switch 116. The physical, mechanical, optical, and / or electrical characteristics of the network port 132 can, but need not, be similar to or identical to those of the connector plug 184, thereby enabling the use of a networking cable 180 having a common interface at both ends thereof.

[0036] Although the system 100 is shown as including four computing nodes 124a-d, it should be appreciated that a greater or lesser number of computing nodes 124a-d can be provided without departing from the scope of the present disclosure. The multi-host NIC 120 is shown as including a number of components that enable the computing nodes 124a-d to share network resources of the TOR switch 116. In other words, the multi-host NIC 120 can be configured to switch data and carry data between each of the computing nodes 124a-d and the network port 132, thereby facilitating communication between the computing nodes 124a-d and the communication network 108. The multi-host NIC 120 is not shown as including the auxiliary card 160 or the CPU 168, while the computing nodes 124a-d are shown as including such features.

[0037] The multi-host NIC 120 is shown as including a number of multi-host ports 136 on a card 128. The card 128 can also include an edge connector 152 that connects with a bus link 156. The bus link 156 can connect with a separate motherboard 140 (or similar physical support substrate) that includes a bus slot 148 and other processing circuitry. In some embodiments, the processing circuitry can be provided as a processor 144 on the motherboard 140. Although not shown, it should be appreciated that the processor 144 can be provided on the card 128, rather than on the separate motherboard 140. Regardless of the physical configuration, the processor 144 can be configured to perform the switching and data aggregation / distribution functions described in connection with the multi-host NIC 120.

[0038] According to at least some embodiments, the processor 144 can correspond to a microprocessor, a central processing unit (CPU), a graphics processing unit (GPU), a data processing unit (DPU), an integrated circuit (IC) chip, a field programmable gate array (FPGA), an application specific integrated circuit (ASIC), combinations thereof, or the like. As will be described in further detail, the processor 144 can be configured to switch data and carry data between one, some, or all of the multi-host ports 136 and the network port 132. For example, the processor 144 can be configured to combine data from the first multi-host port 136, the second multi-host port 136, the third multi-host port 136, and / or the fourth multi-host port 136 and transmit the data through the network port 132. Likewise, the processor 144 can be configured to distribute data arriving from the network port 132 among the first multi-host port 136, the second multi-host port 136, the third multi-host port 136, and / or the fourth multi-host port 136. Thus, the processor 144 can be configured to synchronize communications between the compute nodes 124a-d and the TOR switch 116.

[0039] Although Figure 1A And Figure 1B While each of the compute nodes 124a-d are illustrated as being connected to the TOR switch 116 through the multi-host NIC 120, it should be understood that one or more of the compute nodes 124a-d need not necessarily be connected with the multi-host NIC 120 or use the features provided thereby. For example, some of the compute nodes 124a-d can be connected to the multi-host NIC 120, while other compute nodes 124a-d can be connected directly to the TOR switch 116 via a different port than the connector plug 184 used to connect the TOR switch 116 with the multi-host NIC 120. In such a configuration, the multi-host NIC 120 would be responsible for managing communications between the TOR switch 116 and the compute nodes 124a-d connected therewith, but not the other compute nodes 124a-d that bypass the multi-host NIC 120.

[0040] In some embodiments, the network port 132 can correspond to a network connector mounted on the card 128. One or more of the multi-host ports 136 can correspond to a peripheral component bus connector socket (e.g., a PCIe connector socket). The multi-host ports 136 can provide an attachment mechanism for peripheral component interconnect cables (e.g., PCIe cables, mini SAS HD cables, etc.). Other socket types can be used for the multi-host ports 136 without using PCIe. For example, the peripheral component interconnect cables 176 and the multi-host ports 136 can be configured to utilize RapidIO, InfiniBand, etc.

[0041] The edge connector 152 of the card 128 can be used to carry data between the ports 132, 136 and the motherboard 140. In some embodiments, the edge connector 152 can connect to a bus link 156 that provides a communication path between the card 128 (and its components) and the motherboard 140 (and its components). In some embodiments, the card 128 and the motherboard 140 can be configured in such a way that the edge connector 152 of the card 128 can be inserted into the bus slot 148 of the motherboard 140. Once inserted, the edge connector 152 makes physical and electrical contact with the bus slot 148 and establishes the bus link 156. Depending on the type of communication protocol used to communicate between the multi-host NIC 120 and the compute nodes 124a-d, the bus link 156 can comprise a PCIe link, an InfiniBand link, a RapidIO link. Alternatively or additionally, because the networking cable 184 can correspond to a different type of cable than the peripheral component networking cable 176, and because the processor 144 can perform certain conversion functions described herein, the bus link 156 can also carry data according to the protocol used by the networking cable 180. In alternative configurations, conversion circuitry can reside in the card 128 that converts data at the card 128 rather than relying on the processor 144 to perform such conversions.

[0042] Figure 1A and Figure 1B Each of the compute nodes 124a-d is further illustrated as including a supplemental card 160, a compute node port 164, a processor 168, and a bus link 172. Although each of the compute nodes 124a-d is depicted as having the same components, it can be appreciated that one or more of the compute nodes 124a-d can have different components and / or configurations than the other compute nodes 124a-d. For example, one of the compute nodes 124a-d can include a CPU for the processor 168, while another of the compute nodes 124a-d can include a GPU, DPU, or other type of processing device for the processor 168. Other variations can be contemplated for the various compute nodes 124a-d contained in the server enclosure 112.

[0043] Illustratively, the supplemental card 160 of the compute node 124 can include the compute node port 164. The compute node port 164 can include a similar physical, electrical, and / or optical structure as one of the multi-host ports 136 provided on the multi-host NIC 120. For example, the compute node port 164 can include a peripheral component bus connector receptacle and / or connector plug. The compute node port 164 can be configured to interface with and terminate the peripheral component interconnect cable 176.

[0044] Although not depicted, auxiliary card 160 may include an edge connector that enables physical and electrical connection between auxiliary card 160 and slot, thereby creating bus link 172 and enabling communication between auxiliary card 160 and processor 168. In some embodiments, bus link 172 may be similar to or equivalent to bus link 156 and may correspond to a PCIe link.

[0045] The auxiliary card 160 may include additional circuitry or components to facilitate communication between the auxiliary card 160 and the multi-host NIC 120. For example, the auxiliary card 160 may include a PCIe retimer unit that enables routing of bus link 172 (e.g., a PCIe bus) from compute node 124 to external peripheral interconnect cable 176. In some embodiments, the retimer unit provided in the auxiliary card 160 may include an amplifier that compensates for signal loss caused by bus link 172 and / or peripheral interconnect cable 176.

[0046] Now for reference Figure 2 Further mechanical details of system 100 will be described in accordance with at least some embodiments of the present disclosure. A multi-host NIC 120 is shown disposed on a substrate that can be mounted (e.g., by screws, fasteners, etc.) to a server housing (e.g., a first server housing 112a).

[0047] As described above, each of the multi-host NIC 120 and compute nodes 124a-d can be housed within a separate server enclosure 112a-e within the server rack 104. Providing the multi-host NIC 120 and each compute node 124a-d on separate server enclosures 112a-e also allows different compute nodes 124a-d to be added to and removed from the server rack 104 with minimal impact on other components of the system 100.

[0048] Now for reference Figure 3 The illustrative method 300 will be described in accordance with at least some embodiments of this disclosure. It should be understood that some steps described herein may be performed in parallel with each other, or may correspond to each other or not. Figure 3 The operations described herein shall be executed in any suitable order.

[0049] Method 300 begins by mounting one or more compute nodes 124a-d onto a separate server enclosure (e.g., server enclosure 112b-e) (step 304). In some embodiments, this step may involve sliding or inserting the compute nodes 124a-d into slots in the enclosure body.

[0050] Method 300 continues by installing the multi-host NIC 120 onto a server enclosure 112 (e.g., server enclosure 112a) (step 308). In some embodiments, this step can involve fastening the multi-host NIC 120 to an appropriate socket of the server enclosure body.

[0051] Method 300 can then continue by connecting the multi-host NIC 120 with an auxiliary card 160 of one or more of the compute nodes 124a-d (step 312). For example, one end of the external peripheral component interconnect cable 176 can be connected to the multi-host port 136 of the multi-host NIC 120, and the other end of the external peripheral component interconnect cable 176 can be connected to the compute node port 164 of the auxiliary card 160.

[0052] After establishing a connection between the compute nodes 124a-d and the multi-host NIC 120, method 300 can further continue by connecting the multi-host NIC 120 with the TOR switch 116 (step 316). In some embodiments, the multi-host NIC 120 can be connected with the TOR switch 116 by connecting a first end of the networking cable 180 to the network port 132 of the multi-host NIC 120, and connecting a second end of the networking cable 180 to a connector plug 184 (e.g., port) of the TOR switch 116.

[0053] The TOR switch 116 can then be connected with the communication network 108 (step 320). If the server rack 104 already houses other operational servers or server enclosures, the TOR switch 116 can already be connected with the communication network 108.

[0054] Once all appropriate connections and cables have been established, the method 300 can continue by enabling data flow (e.g., packet transmission) between the multi-host NIC 120 and the auxiliary cards 160 of the various compute nodes 124a-d (step 324). Communication between the multi-host NIC 120 and the compute nodes 124a-d can be facilitated by a first communication protocol (e.g., PCIe, InfiniBand, RapidIO, etc.). The method 300 can also include enabling data flow between the TOR switch 116 and the multi-host NIC 120 to facilitate communication between the compute nodes 124a-d and the communication network 108 (step 328). In some embodiments, a different communication protocol can be used to facilitate communication between the TOR switch 116 and the multi-host NIC 120 than is used between the compute nodes 124a-d and the multi-host NIC 120. For example, Ethernet cables can be used to connect the multi-host NIC 120 with the TOR switch 116, and a communication protocol other than PCIe, InfiniBand, or RapidIO can be used between the multi-host NIC 120 and the TOR switch 116 and / or between the TOR switch 116 and the communication network 108.

[0055] In some embodiments, the steps 324 and 328 can include enabling the multi-host NIC 120 to switch data between the multi-host ports 136 and the communication network 108. The steps can alternatively or additionally include enabling the multi-host NIC 120 to combine data arriving from multiple multi-host ports 136 and transmit the combined data through the network port 132. Alternatively or additionally, the steps can include enabling the multi-host NIC 120 to distribute data arriving from the network port 132 among the various multi-host ports 136.

[0056] In the description, specific details are set forth in order to provide a thorough understanding of embodiments. However, one skilled in the relevant art will recognize that embodiments can be practiced without one or more of the specific details, or with other methods, components, materials, and so forth. In other instances, well-known circuits, processes, algorithms, structures, and techniques have not been shown in detail in order to avoid obscuring an understanding of this description.

[0057] Although the illustrative embodiments of the disclosure have been described herein with reference to the accompanying drawings, it is to be understood that the concepts of the present disclosure can be embodied in other ways and by other means and that the appended claims are intended to cover such and all other modifications and equivalents as fall within the scope of the present disclosure.

Claims

1. A multi-host networking system, comprising: A first computing node, the first computing node comprising: First processing unit; The first compute node port; and A first peripheral component interconnect bus is configured to carry data between the first processing unit and the first computing node port; A multi-host network interface controller, the multi-host network interface controller comprising: A first multi-host port, wherein the first multi-host port is configured to be connected to the first compute node port via a first peripheral interconnect cable; A network port, wherein the network port is configured as a network interface to receive a networking cable; Second multi-host port; and Processing circuitry, configured to convert and carry data between the first multi-host port and the network port; and The second computing node includes: Second processing unit; A second compute node port, wherein the second multi-host port is configured to connect to the second compute node port via a second peripheral interconnect cable; and A second peripheral component interconnect bus is configured to carry data between the second processing unit and the second computing node port, wherein the processing circuitry is configured to convert and carry data between the second multi-host port and the network port, and wherein the processing circuitry is further configured to combine data arriving from the first multi-host port and the second multi-host port, and to transmit the combined data through the network port.

2. The multi-host networking system of claim 1, wherein the processing circuit is configured to convert and carry data between the second multi-host port and the network port, and wherein the processing circuit is further configured to distribute data arriving from the network port between the first multi-host port and the second multi-host port.

3. The multi-host networking system according to claim 1 further includes: A top-of-rack switch configured to communicate with the first compute node and the second compute node via the multi-host network interface controller.

4. The multi-host networking system according to claim 3, wherein the top-of-rack switch includes a top-of-rack network port, the top-of-rack network port being configured to receive a second network interface of the networking cable, and wherein communication between the top-of-rack switch and the multi-host network interface controller is carried by the networking cable.

5. The multi-host networking system according to claim 3, wherein the rack-top switch is further connected to a packet-based communication network.

6. The multi-host networking system according to claim 1, wherein the first computing node further includes an auxiliary card, wherein the auxiliary card includes the first computing node port, and wherein the auxiliary card is connected to the first peripheral component interconnect bus via a peripheral component interconnect bus edge connector.

7. The multi-host networking system according to claim 6, wherein the auxiliary card further includes a retimer unit, the retimer unit routing the first peripheral component interconnect bus from the first computing node to the first computing node port.

8. The multi-host networking system according to claim 7, wherein the retimer unit includes an amplifier that compensates for signal loss caused by the first peripheral component interconnect bus or the first peripheral component interconnect cable.

9. The multi-host networking system according to claim 1, wherein the first processing unit includes a central processing unit, and wherein the first computing node and the multi-host network interface controller are disposed on a separate server shell.

10. A multi-host networking system, comprising: A first server enclosure, the first server enclosure including a first computing node, wherein the first computing node includes: First auxiliary card; First processing unit; The first compute node port; and A first peripheral component interconnect bus is configured to carry data between the first processing unit and the first computing node port; A second server enclosure, the second server enclosure including a multi-host network interface controller, wherein the multi-host network interface controller includes: A first multi-host port, wherein the first multi-host port is configured to be connected to the first compute node port via a first peripheral interconnect cable; A network port, wherein the network port is configured as a network interface to receive a networking cable; Second multi-host port; and Processing circuitry, configured to convert and carry data between the first multi-host port and the network port; and A third server enclosure, the third server enclosure including a second computing node, wherein the second computing node includes: Second auxiliary card; Second processing unit; A second compute node port, wherein the second multi-host port is configured to connect to the second compute node port via a second peripheral interconnect cable; and A second peripheral component interconnect bus is configured to carry data between the second processing unit and the second computing node port, wherein the processing circuitry is configured to convert and carry data between the second multi-host port and the network port, and wherein the processing circuitry is further configured to combine data arriving from the first multi-host port and the second multi-host port, and transmit the data through the network port.

11. The multi-host networking system according to claim 10, wherein the first auxiliary card includes the first computing node port, and wherein the second auxiliary card includes the second computing node port.

12. The multi-host networking system of claim 10, wherein the processing circuitry is configured to convert and carry data between the second multi-host port and the network port, and wherein the processing circuitry is further configured to distribute data arriving from the network port between the first multi-host port and the second multi-host port.

13. The multi-host networking system according to claim 10, wherein the first auxiliary card includes a retimer unit, the retimer unit routing the first peripheral component interconnect bus from the first computing node to the first computing node port.

14. The multi-host networking system of claim 13, wherein the retimer unit includes an amplifier that compensates for signal loss caused by the first peripheral component interconnect bus or the first peripheral component interconnect cable.

15. A method for networking multiple hosts, comprising: A multi-host network interface controller is mounted on a first server enclosure. The multi-host network interface controller includes a first multi-host port, which is configured to connect to a first compute node port of a first compute node via a first peripheral network cable. The multi-host network interface controller also includes a network port, which is configured to receive the network cable. The multi-host network interface controller further includes processing circuitry configured to convert and carry data between the first multi-host port and the network port. A first auxiliary card is installed on a second server housing, wherein the first auxiliary card includes the first computing node, and wherein the first auxiliary card includes a first peripheral component interconnect bus, the first peripheral component interconnect bus being configured to carry data between a first processing unit and a port of the first computing node; A data stream is enabled between the first auxiliary card and the multi-host network interface controller via a first communication protocol; A data stream is enabled between the multi-host network interface controller and the communication network via a second communication protocol, thereby facilitating communication between the first auxiliary card and the communication network. The second auxiliary card is installed on the third server housing, wherein the second auxiliary card includes a second computing node, and wherein the second auxiliary card includes a second peripheral component interconnect bus, the second peripheral component interconnect bus being configured to carry data between the second processing unit and the second computing node port; Connect the second computing node port to the second multi-host port of the multi-host network interface controller; This enables the multi-host network interface controller to convert and carry data between the second multi-host port and the communication network; This enables the multi-host network interface controller to combine data arriving from the first multi-host port and the second multi-host port, and to transmit the combined data through the network port; as well as This enables the multi-host network interface controller to (i) distribute data arriving from the network port between the first multi-host port and the second multi-host port and / or (ii) combine data arriving from the first multi-host port and the second multi-host port for transmission through the network port.

Images