Network interface card management system and server

Abstract

The present application discloses a management system and a server for a network interface card, and relates to the field of server technology. By connecting a management module to a network interface card, the network interface card is powered when the server is powered on, and the management module switches the connection status of each server computing node and at least one network interface card according to a management signal sent by each server computing node, the control and power-on of the network interface card are migrated from the server computing node to an independent management module, thereby realizing unified management at the hardware level. This can, to a certain extent, avoid the problem of network interface card resources being wasted due to the inability to power the network interface card when the server computing node is powered off, and the problem of the inability to switch the NCSI function of the server computing node, thereby improving the stability of the server and the resource utilization efficiency.

Description

Technical Field

[0001] The present application relates to the technical field of servers, and in particular to a network interface card management system and a server. Background Art

[0002] With the rapid development of cloud computing, artificial intelligence, and high-performance computing technologies, multi-node, high-density server architectures have become essential data center infrastructure. These systems integrate 4-8 compute nodes in a single 2U chassis to collaboratively process large-scale tasks and achieve high-density computing. However, in actual engineering deployments of multi-node, high-density servers, the server compute nodes typically control and power the network interface cards (NICs). Consequently, when the NICs are in multi-host mode, the power supply to the NICs on multiple nodes cannot be coordinated, resulting in wasted NIC resources. Furthermore, the NCSI (Network Controller Sideband Interface) function of the server compute nodes cannot be switched. Summary of the Invention

[0003] The present application provides a network interface card management system and server to at least solve the problem in the related art that when the network interface card is in Multihost mode, the power supply of multiple node network interface cards cannot be coordinated, resulting in waste of network interface card resources, and the problem that the NCSI function of the server computing node cannot be switched.

[0004] The present application provides a network interface card management system, comprising: a management module and multiple server computing nodes, each server computing node being connected to the management module, and the management module being connected to at least one network interface card, wherein the management module is configured to supply power to the at least one network interface card when the server is powered on; the multiple server computing nodes are configured to send management signals of corresponding server computing nodes to the management module; the management module is further configured to receive the management signal and, upon determining that at least one network interface card is in a target operating mode, switch the connection status of each server computing node with the at least one network interface card according to the management signal.

[0005] This application connects the management module to the network interface card, powers the network interface card when the server is powered on, and the management module switches the connection status of each server computing node and at least one network interface card according to the management signal sent by each server computing node. In this way, the control and power-on of the network interface card are migrated from the server computing node to an independent management module, thereby realizing unified management at the hardware level. To a certain extent, it can avoid the problem of network interface card resources being wasted due to the inability to power the network interface card when the server computing node is powered off, and the problem of being unable to switch the NCSI function of the server computing node, thereby improving the stability of the server and resource utilization efficiency. BRIEF DESCRIPTION OF THE DRAWINGS

[0006] In order to more clearly illustrate the embodiments of the present application, the following is a brief introduction to the drawings required for use in the embodiments. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without any creative work.

[0007] Figure 1 A schematic structural diagram of a network interface card management system according to the related art of some embodiments of the present application;

[0008] Figure 2 is a schematic structural diagram of a network interface card management system according to some embodiments of the present application;

[0009] Figure 3 is a structural diagram of a network interface card management system according to some other embodiments of the present application;

[0010] Figure 4 is a structural diagram of a network interface card management system according to some other embodiments of the present application;

[0011] Figure 5 Flowchart of the process of powering on and off a network interface card and switching the connection status between a server computing node and the network interface card according to some embodiments of the present application;

[0012] Figure 6 Schematic diagram of a server according to some embodiments of the present application. DETAILED DESCRIPTION

[0013] The following will be combined with the accompanying drawings in the embodiments of this application to clearly and completely describe the technical solutions in the embodiments of this application. Obviously, the embodiments described are only part of the embodiments of this application, not all of them. Based on the embodiments in this application, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of this application.

[0014] It should be noted that, in the description of this application, the terms "comprises," "includes," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or device comprising a series of elements includes not only those elements, but also other elements not explicitly listed, or elements inherent to such process, method, article, or device. The terms "first," "second," etc., in this application are used to distinguish similar objects, and are not used to describe a particular order or sequence.

[0015] In order to enable those skilled in the art to better understand the present application, the present application is further described in detail below with reference to the accompanying drawings and specific implementation methods.

[0016] The following describes in detail the network interface card management system and server according to the embodiments of the present application with reference to the accompanying drawings.

[0017] In the related art, refer to Figure 1 The management system of the network interface card includes a management module and multiple server computing nodes. In the following description, the management system of the network interface card includes two server computing nodes, such as server computing node A and server computing node B, but this is not a limitation of this application.

[0018] Server compute node A includes a central processing unit (CPUa), a motherboard, a node control unit, and a power supply interface (not shown). The motherboard can be a motherboard complex programmable logic device (MB CPLD), and the node control unit can be a baseboard management controller (BMC). Server compute node B includes CPUb. The management module includes a management unit, which can be a main processor complex programmable logic device (MP CPLD).

[0019] The management unit of the management module is connected to server compute node A and server compute node B, respectively. The power supply interface of server compute node A is connected to the network interface card, directly powering the network interface card. The mainboard of server compute node A is connected to the network interface card. The mainboard is configured to collect the power management signal from CPUa on server compute node A, determine a power control signal based on the power management signal, and send the power control signal to the network interface card. The network interface card determines the operating mode based on the power control signal. For example, if the power management signal from CPUa is the first power management signal (CPUa is powered on), the power control signal from the network interface card is determined to be the first source control signal. Upon receiving the first source control signal, the network interface card determines the operating mode to be the main power mode, meaning that all network interface card functions are enabled. If the power management signal from CPUa is the second power management signal (CPUa is powered off), the power control signal from the network interface card is determined to be the second source control signal. Upon receiving the second source control signal, the network interface card determines the operating mode to be the auxiliary power mode, meaning that some network interface card functions are enabled. The network interface card (NIC) connects to CPUa of server compute node A and the CPU of server compute node B through Peripheral Component Interconnect Express (PCIE) lines, forming a 2x8 PCIE topology. The NIC connects to the BMC of server compute node A to implement the NCSI function. The NIC can be an OCP (Open Compute Project) NIC.

[0020] Depend on Figure 1 It can be seen that the two nodes are combined into a Mulihost, and the power supply of the network interface card is controlled only by the server computing node A. The network interface card is connected to the CPU of the two nodes through the PCIE line to form a 2x8 PCIE topology and receives the NCSI bus of the node control unit of the server computing node A.

[0021] When two nodes are combined into a multihost, the CPU power on / off status of server compute node A and server compute node B must be synchronized. Otherwise, if only server compute node A is shut down, the motherboard of server compute node A will not be able to power the network interface card. As a result, server compute node B will be powered on but unable to connect to the network interface card, resulting in a loss of functionality. If only server compute node B is shut down, the network interface card will lose the PCIE line connection with server compute node B and can only support a 1x8 connection with server compute node A, resulting in a waste of resources.

[0022] In addition, since the network interface card cannot switch the NCSI function of the two server computing nodes by itself, when the two nodes are currently connected to the network interface card, server computing node A can implement the NCSI function with the network interface card, but server computing node B cannot, which will also cause functional loss.

[0023] Based on this, refer to Figure 2 The network interface card management system 100 includes: a management module 11 and multiple server computing nodes 12, each server computing node 12 is respectively connected to the management module 11, and the management module 11 is connected to at least one network interface card 200, wherein the management module 11 is used to power the at least one network interface card 200 when the server is powered on; the multiple server computing nodes 12 are used to send corresponding server computing node management signals to the management module 11; the management module 11 is also used to receive the management signal, and when it is determined that at least one network interface card 200 is in the target operation mode, switch the connection state between each server computing node 12 and the at least one network interface card 200 according to the management signal.

[0024] In the embodiment of the present application, the example in which the multiple server computing nodes 12 include server computing node A and server computing node B is used for illustration, but this is not intended to limit the present application.

[0025] Specifically, the management module 11 is connected to at least one network interface card 200 and is configured to provide power to all network interface cards when the server is powered on by AC power, for example, providing 3.3V and 12V voltages.

[0026] The server computing node A and the server computing node B are respectively connected to the management module 11. After the server AC is powered on, the server computing node A and the server computing node B respectively send corresponding management signals to the management module 11, such as sending power management signals. That is, the server computing node A sends a power management signal a to the management module 11, and the server computing node B sends a power management signal b to the management module 11. The management module 11 receives and determines the operating mode of at least one network interface card 200 based on the power management signal a and the power management signal b. For example, when one of the power management signal a and the power management signal b is a server computing node power-on signal, the operating mode of at least one network interface card 200 is determined to be the target operating mode, where the target operating mode represents the main power mode of the network interface card, that is, all the functions of the network interface card are online.

[0027] When determining that at least one network interface card 200 is in the target operating mode, the management module 11 continues to receive management signals, such as power management signals and interface management signals, and switches the connection status between each server computing node 12 and the at least one network interface card 200 according to the power management signals and the interface management signals. For example, the management module 11 may switch the PCIE line connection between each server computing node 12 and the at least one network interface card 200 according to the power management signal, and the management module 11 may switch the NCSI connection between each server computing node 12 and the at least one network interface card 200 according to the interface management signal.

[0028] It should be noted that the number of network interface cards 200 may be one or more. One network interface card 200 can provide network services for multiple server computing nodes 12, and multiple network interface cards can also provide network services for one server computing node 12. For example, when there is only one network interface card 200, for example, network interface card C, when one of power management signal a and power management signal b is a server computing node startup signal, the operating mode of network interface card C can be determined to be the target operating mode; when there are multiple network interface cards 200, for example, network interface card C and network interface card D, when one of power management signal a and power management signal b is a server computing node startup signal, the operating modes of network interface card C and network interface card D can be determined to be the target operating mode.

[0029] Furthermore, one network interface card 200 can establish PCIE line connections for multiple server computing nodes 12, and multiple network interface cards 200 can also establish PCIE line connections for one server computing node. For example, when there is only one network interface card 200, such as network interface card C, the management module 11 switches the PCIE line connections between server computing node A, server computing node B, or server computing node A and server computing node B and network interface card C according to the power management signal; when there are multiple network interface cards 200, such as network interface card C and network interface card D, the management module 11 switches the PCIE line connections between server computing node A, server computing node B, or server computing node A and server computing node B and network interface card C and network interface card D according to the power management signal.

[0030] Furthermore, one network interface card 200 can only provide NCSI service for one server computing node 12. Exemplarily, when the number of network interface cards 200 is one, such as network interface card C, the default server computing node A establishes an NCSI connection with the network interface card C, and the management module 11 can switch the NCSI connection between the server computing node B and the network interface card C according to the interface management signal sent by the server computing node B; when the number of network interface cards 200 is multiple, such as including network interface card C and network interface card D, the default server computing node A establishes an NCSI connection with the network interface card C, the management module 11 can switch the NCSI connection between the server computing node B and the network interface card C according to the interface management signal sent by the server computing node B, and can also switch the NCSI connection between the server computing node B and the network interface card D according to the interface management signal sent by the server computing node B.

[0031] This application connects the management module to the network interface card, powers the network interface card when the server is powered on, and the management module switches the connection status of each server computing node and at least one network interface card according to the management signal sent by each server computing node. In this way, the control and power-on of the network interface card are migrated from the server computing node to an independent management module, thereby realizing unified management at the hardware level. To a certain extent, it can avoid the problem of network interface card resources being wasted due to the inability to power the network interface card when the server computing node is powered off, and the problem of being unable to switch the NCSI function of the server computing node, thereby improving the stability of the server and resource utilization efficiency.

[0032] In some embodiments, reference Figure 3 Each server computing node 12 includes a corresponding node control unit 121 and a central processing unit 122. The management module 11 includes a management unit 111, a power supply interface (not shown) and a bus switching unit 112.

[0033] The power supply interface is connected to at least one network interface card 200, each node control unit 121 is connected to the management unit 111, the management unit 111 is connected to at least one network interface card 200, each central processing unit 122 is connected to the bus switching unit 112, the bus switching unit 112 is connected to at least one network interface card 200, and the management unit 111 is connected to the bus switching unit 112, wherein the bus switching unit 112 can be a PCIE Switch chip, and no specific restrictions are made here.

[0034] The management module 11 is used to control the power supply interface to supply power to at least one network interface card 200 when the server is powered on; multiple node control units 121 are used to collect power management signals of corresponding server computing nodes and send corresponding power management signals to the management unit 111; the management unit 111 is used to receive the power management signal and, when determining that at least one network interface card 200 is in the target operating mode, control the bus switching unit 112 to switch the connection status between each central processing unit 122 and at least one network interface card 200 according to each power management signal.

[0035] In the embodiment of the present application, the example in which server computing node A includes node control unit a and central processing unit a, and server computing node B includes node control unit b and central processing unit b is used for illustration, but this is not a limitation of the present application.

[0036] Specifically, the power supply interface of the management module 11 is connected to at least one network interface card 200 and is used to power all network interface cards when the server is powered on by AC, for example, providing 3.3V and 12V voltages. Each node control unit 121 is connected to the management unit 111 via an I2C (Inter-Integrated Circuit, a serial bus protocol for short-distance communication) serial data line, and the management unit 111 is connected to at least one network interface card 200 via an I2C serial data line. Each central processing unit 122 is connected to the bus switching unit 112 via a PCIE bus, and the bus switching unit 112 is connected to at least one network interface card 200 via a PCIE bus. The management unit 111 is in communication with the bus switching unit 112.

[0037] Server computing node A includes a node control unit a and a central processing unit a, and server computing node B includes a node control unit b and a central processing unit b. Node control unit a is used to collect power management signal a from server computing node A. Node control unit b is used to collect power management signal b from server computing node B. After the server AC power is applied, node control units a and b respectively collect power management signal a and power management signal b and send them to management unit 111. Management unit 111 determines whether the network interface card is in the target operating mode based on power management signal a and power management signal b. For example, if one or both of power management signal a and power management signal b are target power management signals (power-on signals), the network interface card is determined to be in the target operating mode.

[0038] After the management unit 111 determines that the network interface card is in the target operating mode, the node control unit a and the node control unit b monitor the power management signal a and the power management signal b in real time and transmit the power management signal a and the power management signal b to the management unit 111. The management unit 111 controls the bus switching unit 112 to switch the connection state between the central processing unit a and the central processing unit b and the network interface card 200 according to the power management signal a and the power management signal b. For example, the management unit 111 controls the bus switching unit 112 to switch the connection state between the central processing unit a and the network interface card 200 to an on state or a off state according to the power management signal a; and the management unit 111 controls the bus switching unit 112 to switch the connection state between the central processing unit b and the network interface card 200 to an on state or a off state according to the power management signal b.

[0039] In some embodiments, the management unit 111 is configured to, when the received power management signal is a first power management signal, control the bus switching unit 112 to switch the connection state between the central processing unit 122 and the at least one network interface card 200 to a conducting state. The management unit 111 is further configured to, when the received power management signal is a second power management signal, control the bus switching unit 112 to switch the connection state between the central processing unit 122 and the at least one network interface card 200 to a disconnected state.

[0040] Specifically, the power management signal a includes a first power management signal a and a second power management signal a, wherein the first power management signal a is used to indicate that the central processing unit a is powered on, and the second power management signal a is used to indicate that the central processing unit a is powered off. The power management signal b includes a first power management signal b and a second power management signal b, wherein the first power management signal b is used to indicate that the central processing unit b is powered on, and the second power management signal b is used to indicate that the central processing unit b is powered off.

[0041] As a specific example, when the management unit 111 is connected to a network interface card (e.g., network interface card C), if the management unit 111 receives the first power management signal a and the first power management signal b, the control bus switching unit 112 switches the connection state between the central processing unit a and the central processing unit b and the network interface card C to the on state, thereby forming a multihost topology; if the management unit 111 receives the second power management signal a and the first power management signal b, the control bus switching unit 112 switches the connection state between the central processing unit a and the network interface card C to the off state, thereby connecting the central processing unit b to the network interface card C. The connection state of the central processing unit a and the network interface card C is switched to the on state, forming a singlehost topology; if the management unit 111 receives the first power management signal a and the second power management signal b, the bus switching unit 112 is controlled to switch the connection state between the central processing unit a and the network interface card C to the on state, and the connection state between the central processing unit b and the network interface card C is switched to the disconnected state, forming a singlehost topology; if the management unit 111 receives the second power management signal a and the second power management signal b, there is no need to control the bus switching unit 112 to switch the connection state between the central processing unit a and the central processing unit b and the network interface card C.

[0042] As another specific example, when the management unit 111 is connected to multiple network interface cards (for example, network interface card C and network interface card D), if the management unit 111 receives the first power management signal a and the first power management signal b, the bus switching unit 112 is controlled to switch the connection state between the central processing unit a and the central processing unit b and the network interface card C and the network interface card D to the conductive state; if the management unit 111 receives the second power management signal a and the first power management signal b, the bus switching unit 112 is controlled to switch the connection state between the central processing unit a and the network interface card C and the network interface card D to the disconnected state. The connection state between the central processing unit b and the network interface card C is switched to the on state; if the management unit 111 receives the first power management signal a and the second power management signal b, the bus switching unit 112 is controlled to switch the connection state between the central processing unit a and the network interface card C and the network interface card D to the on state, and the connection state between the central processing unit b and the network interface card C is switched to the disconnected state; if the management unit 111 receives the second power management signal a and the second power management signal b, there is no need to control the bus switching unit 112 to switch the connection state between the central processing unit a and the central processing unit b and the network interface card C and the network interface card D.

[0043] In this way, the management module directly supplies power to the server computing nodes, achieving the disconnection between the network interface card timing and the power on / off status of each server computing node; the management unit reads the power management signal of the server computing node and controls the bus switching unit to switch the connection status between the central processing unit and the network interface card according to the power management signal, thereby achieving the most rational network topology and reducing resource waste.

[0044] In some embodiments, reference Figure 4 The network interface card management system 100 includes: a plurality of server computing nodes 12 including a plurality of first server computing nodes and a plurality of second server computing nodes, wherein the first server computing nodes represent server computing nodes in which the connection state between the node control unit 121 and the network interface card 200 is in a conducting state, and the second server computing nodes represent server computing nodes in which the connection state between the node control unit 121 and the same network interface card 200 is in a disconnected state.

[0045] The management module 11 includes an interface switching unit 113 and a management unit 111. The first node control unit of the first server computing node and the second node control unit of the second server computing node are respectively connected to the interface switching unit 113. The interface switching unit 113 is connected to at least one network interface card 200. The management unit 111 is connected to the interface switching unit 113. The interface switching unit 113 can be an NCSI Switch chip, which is not specifically limited here.

[0046] The second node control unit is used to send an interface management signal to the management unit 111; the first node control unit is used to receive the interface management signal sent by the management unit 111 and send a response to the interface management signal to the management unit 111; the management unit 111 is used to receive the interface management signal sent by the second node control unit when it is determined that at least one network interface card 200 is in the target operation mode, and send the interface management signal to the first node control unit, and receive a response to the interface management signal sent by the first node control unit; the management unit 111 is also used to control the interface switching unit 113 to switch the connection status between the second node control unit and at least one network interface card 200 according to the response to the interface management signal.

[0047] Specifically, in an embodiment of the present application, it is assumed that the connection state of the node control unit a and the network interface card 200 is in the on state, and the connection state of the node control unit b and the same network interface card 200 is in the off state. That is to say, the server computing node A is the first server computing node, and the server computing node B is the second server computing node. The node control unit a of the server computing node A and the node control unit b of the server computing node B are respectively connected to the interface switching unit 113 through the NCSI bus, and the interface switching unit 113 is connected to at least one network interface card 200 through the NCSI bus, and the management unit 111 is communicated with the interface switching unit 113.

[0048] When node control unit b needs NCSI service, management unit 111 needs to determine whether the network interface card is in the target working mode based on the power management signals collected by node control unit a and node control unit b. The specific process has been described in detail in the above embodiment and will not be repeated here.

[0049] After the management unit 111 determines that the network interface card is in the target operation mode, the node control unit b sends an interface management signal (NCSI service request signal) to the management unit 111. Since the node control unit b and the node control unit a require the same network interface card (for example, network interface card C) to provide NCSI services, the management unit 111 needs to forward the interface management signal to the node control unit a after receiving the interface management signal. The node control unit a sends a response to the interface management signal to the management unit 111 based on whether it requires the NCSI service. The management unit 111 controls the interface switching unit 113 to switch the connection state between the node control unit b and the network interface card C based on the response to the management signal. For example, the management unit 111 controls the interface switching unit 113 to switch the connection state between the node control unit b and the network interface card C to an on state or a disconnected state based on the response to the management signal.

[0050] In some embodiments, the management unit 111 is configured to, when the response to the interface management signal received is a first response, control the interface switching unit 113 to switch the connection state between the second node control unit and the at least one network interface card 200 to an on state. The management unit 111 is further configured to, when the response to the interface management signal received is a second response, control the interface switching unit 113 to switch the connection state between the second node control unit and the at least one network interface card 200 to a off state. The first node control unit is configured to send a first response to the management unit 111 when the first node control unit does not require the network controller sideband interface service. The first node control unit is further configured to send a second response to the management unit 111 when the first node control unit does require the network controller sideband interface service.

[0051] Specifically, after receiving the interface management signal, node control unit a determines whether it needs NCSI service. If it is determined that node control unit a does not need NCSI service, it sends a first response to the management unit 111; if it is determined that node control unit a still needs NCSI service, it sends a second response to the management unit 111.

[0052] As a specific example, when the management unit 111 is connected to a network interface card (for example, network interface card C), network interface card C needs to provide NCSI services for node control unit a and node control unit b. If node control unit a sends a first response to the management unit 111, indicating that node control unit a does not need NCSI services, the management unit 111 controls the interface switching unit 113 to switch the connection state between node control unit b and network interface card C to a conducting state, so that network interface card C provides NCSI services for node control unit b; if node control unit a sends a second response to the management unit 111, indicating that node control unit a still needs NCSI services, the management unit 111 controls the interface switching unit 113 to switch the connection state between node control unit b and network interface card C to a disconnected state, so that network interface card C continues to provide NCSI services for node control unit a.

[0053] As another specific example, when the management unit 111 is connected to multiple network interface cards (for example, network interface card C and network interface card D), if network interface card C is still needed to provide NCSI services for node control unit a and node control unit b, then as in the above embodiment, the management unit 111 needs to control the interface switching unit 113 to switch the connection status of node control unit b and network interface card C according to the response to the interface management signal sent by node control unit a, which is not repeated here; if network interface card C needs to provide NCSI services for node control unit a, and network interface card D needs to provide NCSI services for node control unit b, that is, node control unit a and node control unit b use different network interface cards to provide NCSI services, then after receiving the interface management signal sent by node control unit b, the management unit 111 controls the interface switching unit 113 to switch the connection status of node control unit b and network interface card D to the on state, so that network interface card D provides NCSI services for node control unit b, while network interface card C still provides NCSI services for node control unit a.

[0054] In this way, the management unit controls the interface switching unit to switch the connection status between the second node control unit and the network interface card based on the connection status between the first node control unit and the network interface card, so that each server node can implement the NCSI function.

[0055] In some embodiments, the management unit 111 is configured to receive a power management signal from each server computing node 12 and determine an operating mode of at least one network interface card 200 according to the power management signal.

[0056] Specifically, node control unit a is configured to collect power management signal a from server compute node A, and node control unit b is configured to collect power management signal b from server compute node B. After the server is powered on via AC, node control units a and b respectively collect power management signal a and b, and send them to management unit 111. Management unit 111 determines the operating mode of at least one network interface card 200 based on power management signal a and b. For example, if one or both of power management signal a and b are target power management signals, network interface card 200 is determined to be in the target operating mode, where the target power management signal refers to the power-on signal of the server compute node, and the target operating mode refers to the network interface card being in the primary power mode. If neither power management signal a nor b is a target power management signal, network interface card 200 is determined to be in the target operating mode, i.e., in the auxiliary power mode.

[0057] After the management unit 111 determines that the network interface card is in the target operating mode, it sends a first control signal to at least one network interface card 200. The at least one network interface card 200 receives the first control signal and determines that the operating mode is the main power mode based on the first control signal. After the management unit 111 determines that the network interface card is not in the target operating mode, it sends a second control signal to the at least one network interface card 200. The at least one network interface card 200 receives the second control signal and determines that the operating mode is the auxiliary power mode based on the second control signal.

[0058] In some embodiments, the management unit 111 is configured to determine that the operation mode is a target operation mode when the power management signal received from at least one server computing node is a target power management signal.

[0059] Specifically, the power management signal a includes a first power management signal a and a second power management signal a, wherein the first power management signal a is used to indicate that the central processing unit a is powered on, and the second power management signal a is used to indicate that the central processing unit a is powered off. The power management signal b includes a first power management signal b and a second power management signal b, wherein the first power management signal b is used to indicate that the central processing unit b is powered on, and the second power management signal b is used to indicate that the central processing unit b is powered off.

[0060] As a specific example, when the management unit 111 is connected to a network interface card (for example, network interface card C), if the management unit 111 receives a first power management signal a and a first power management signal b, it determines that the operating mode of the network interface card C is the target operating mode; if the management unit 111 receives a second power management signal a and a first power management signal b, it determines that the operating mode of the network interface card C is the target operating mode; if the management unit 111 receives the first power management signal a and the second power management signal b, it determines that the operating mode of the network interface card C is the target operating mode; if the management unit 111 receives the second power management signal a and the second power management signal b, it determines that the operating mode of the network interface card C is not the target operating mode.

[0061] As another specific example, when the management unit 111 is connected to multiple network interface cards (for example, network interface card C and network interface card D), if the management unit 111 receives a first power management signal a and a first power management signal b, it determines that the operating mode of network interface card C and network interface card D is the target operating mode; if the management unit 111 receives a second power management signal a and a first power management signal b, it determines that the operating mode of network interface card C and network interface card D is the target operating mode; if the management unit 111 receives the first power management signal a and the second power management signal b, it determines that the operating mode of network interface card C and network interface card D is the target operating mode; if the management unit 111 receives the second power management signal a and the second power management signal b, it determines that the operating mode of network interface card C and network interface card D is not the target operating mode.

[0062] In this way, this application migrates all network interface cards from the server computing nodes to the management module, which is controlled by the management module. The management unit of the management module controls the operating mode of the network interface card by monitoring the power on and off conditions of the server computing nodes, thereby achieving disconnection between the network interface card timing and the power on and off status of each server computing node.

[0063] In some embodiments, at least one network interface card 200 includes a storage unit, the management module 11 includes a management unit 111, and the management unit 111 is connected to the storage unit (not shown), wherein the management unit 111 is used to read the field replaceable unit information in the storage unit after the server is powered on, and control the server according to the field replaceable unit information.

[0064] Specifically, after the management module 11 supplies power to all network interface cards, all network interface cards enter the identification mode. At this time, the management unit 111 reads the storage units of all network interface cards through the I2C serial data line to obtain the FRU (Field Replaceable Unit) information of each network interface card and controls the server according to the FRU information.

[0065] For example, the FRU information includes the mapping relationship between the temperature of the network interface card and the server fan speed and the fast bus topology connection relationship of the network interface card's peripheral components interconnected. The server fan speed can be determined based on the mapping relationship between the temperature of the network interface card and the server fan speed, and the server can be controlled based on the server fan speed; the number of server computing nodes supported by the network interface card can also be determined based on the fast bus topology connection relationship of the network interface card's peripheral components interconnected, and the network interface card 200 can be controlled to establish a connection with the server computing node 12 based on the number of server computing nodes supported by the network interface card.

[0066] In some embodiments, the field replaceable unit information includes a mapping relationship between the temperature of the network interface card and the speed of the server fan, wherein the management unit 111 is used to determine the target speed of the server fan based on the temperature of the network interface card and the mapping relationship between the temperature of the network interface card and the speed of the server fan, and control the server fan to run at the target speed.

[0067] For example, when the management unit 111 is connected to a network interface card (e.g., network interface card C), the temperature of the network interface card C can be detected and acquired by a temperature sensor provided at the network interface card C, and the temperature of the network interface card C is transmitted to the management unit 111. The management unit 111 determines the server fan speed corresponding to the temperature of the network interface card C based on the temperature of the network interface card C and a pre-read mapping relationship between the temperature of the network interface card and the server fan speed, and controls the server fan to run at this speed.

[0068] When the management unit 111 is connected to multiple network interface cards (for example, network interface card C and network interface card D), the temperature of network interface card C can be detected and obtained by a temperature sensor set at network interface card C, and the temperature of network interface card D can be detected and obtained by a temperature sensor set at network interface card D. The temperatures of network interface card C and network interface card D are transmitted to the management unit 111, and the management unit 111 calculates the average temperature of the two. Based on the temperature average and a pre-read mapping relationship between the temperature of the network interface card and the server fan speed, the management unit 111 determines the server fan speed corresponding to the temperature average, and controls the server fan to run at this speed.

[0069] In some embodiments, the field replaceable unit information also includes a peripheral component interconnection fast bus topology connection relationship of the network interface card, and the peripheral component interconnection fast bus topology connection relationship of the network interface card represents the number of server computing nodes supported by the network interface card, wherein the management unit 111 is used to control the network interface card to establish a connection with the server computing node according to the peripheral component interconnection fast bus topology connection relationship of the network interface card.

[0070] Specifically, after acquiring the number of server computing nodes supported by the network interface card, the management unit 111 controls the network interface card to establish connections with the server computing nodes according to the number of server computing nodes supported by the network interface card.

[0071] Through the description of the above implementation methods, those skilled in the art can clearly understand that the method according to the above embodiment can be implemented by means of software plus the necessary general hardware platform, and of course it can also be implemented by hardware, but in many cases the former is a better implementation method.

[0072] As a specific example, see Figure 5 The process of powering on and off the network interface card and switching the connection status between the server computing node and the network interface card includes the following steps:

[0073] S301: The server is not powered on and the management unit is not started.

[0074] S302: The server is powered on by AC, and the management unit starts and obtains the status of the server computing nodes and network interface cards.

[0075] S303: The network interface card enters the identification mode, and only the storage unit and the scanning link device are powered.

[0076] After the management module 11 supplies power to all network interface cards, all network interface cards enter identification mode. At this time, the management unit 111 reads the storage units of all network interface cards through the I2C serial data line to obtain the FRU (Field Replaceable Unit) information of each network interface card and controls the server according to the FRU information.

[0077] For example, the FRU information includes the mapping relationship between the temperature of the network interface card and the server fan speed and the fast bus topology connection relationship of the network interface card's peripheral components interconnected. The server fan speed can be determined based on the mapping relationship between the temperature of the network interface card and the server fan speed, and the server can be controlled based on the server fan speed; the number of server computing nodes supported by the network interface card can also be determined based on the fast bus topology connection relationship of the network interface card's peripheral components interconnected, and the network interface card can be controlled to establish a connection with the server computing nodes based on the number of server computing nodes supported by the network interface card.

[0078] S304: The management unit reads the power management signal of the server computing node and determines that no node is currently powered on.

[0079] S305, the network interface card enters the auxiliary power mode.

[0080] For example, when neither the power management signal a nor the power management signal b is a target power management signal, it is determined that the network interface card is not in the target operation mode, that is, it is determined that the network interface card is in the auxiliary power mode.

[0081] S306: The node is powered on.

[0082] S307, the network interface card enters the main power mode.

[0083] For example, when one or both of the power management signal a and the power management signal b are target power management signals, it is determined that the network interface card is in the target operation mode, that is, it is determined that the network interface card is in the main power mode.

[0084] S308 , the management unit monitors the power management signal and the interface management signal of each node in real time.

[0085] S309: A node is shut down or unplugged.

[0086] S310: The management unit controls the bus switching unit to switch the connection state between the central processing unit and the network interface card according to the power management signal of the node.

[0087] After the management unit 111 determines that the network interface card is in the target operating mode, the node control unit a and the node control unit b monitor the power management signal a and the power management signal b in real time and transmit the power management signal a and the power management signal b to the management unit 111. The management unit 111 controls the bus switching unit 112 to switch the connection state between the central processing unit a and the central processing unit b and the network interface card according to the power management signal a and the power management signal b. For example, the management unit 111 controls the bus switching unit 112 to switch the connection state between the central processing unit a and the network interface card to an on state or a off state according to the power management signal a; and the management unit 111 controls the bus switching unit 112 to switch the connection state between the central processing unit b and the network interface card to an on state or a off state according to the power management signal b.

[0088] S311: All nodes are shut down and no network interface card is required.

[0089] S312: The management unit reads the connection status between the first node control unit and the network interface card.

[0090] S313: If the second node control unit needs to switch the network controller sideband interface service, the management unit controls the interface switching unit to switch the connection status between the second node control unit and the network interface card based on the connection status between the first node control unit and the network interface card.

[0091] For example, assuming that the connection state of node control unit a and network interface card is in the on state, the connection state of node control unit b and the same network interface card is in the off state. That is to say, server computing node A is the first server computing node, and server computing node B is the second server computing node. The node control unit a of server computing node A and the node control unit b of server computing node B are respectively connected to the interface switching unit 113 through the NCSI bus, the interface switching unit 113 is connected to at least one network interface card 200 through the NCSI bus, and the management unit 111 is communicated with the interface switching unit 113.

[0092] When node control unit b needs NCSI service, management unit 111 needs to determine whether the network interface card is in the target working mode based on the power management signals collected by node control unit a and node control unit b. The specific process has been described in detail in the above embodiment and will not be repeated here.

[0093] After the management unit 111 determines that the network interface card is in the target operation mode, the node control unit b sends an interface management signal (NCSI service request signal) to the management unit 111. Since the node control unit b and the node control unit a require the same network interface card to provide the NCSI service, the management unit 111 needs to forward the interface management signal to the node control unit a after receiving the interface management signal. The node control unit a sends a response to the interface management signal to the management unit 111 based on whether it needs the NCSI service. The management unit 111 controls the interface switching unit 113 to switch the connection state between the node control unit b and the network interface card based on the response to the management signal. For example, the management unit 111 controls the interface switching unit 113 to switch the connection state between the node control unit b and the network interface card to an on state or a off state based on the response to the management signal.

[0094] To summarize, the present application connects the management module to the network interface card, powers the network interface card when the server is powered on, and the management module switches the connection status of each server computing node and at least one network interface card according to the management signal sent by each server computing node. In this way, the control and power-on of the network interface card are migrated from the server computing node to an independent management module, thereby realizing unified management at the hardware level. This can, to a certain extent, avoid the problem of waste of network interface card resources caused by the inability to power the network interface card when the server computing node is powered off, and the problem of inability to switch the NCSI function of the server computing node, thereby improving the stability of the server and resource utilization efficiency.

[0095] Reference Figure 6 , an embodiment of the present application further provides a server 10, including the aforementioned network interface card management system 100.

[0096] Professionals may further appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, computer software, or a combination of the two. In order to clearly illustrate the interchangeability of hardware and software, the above description has generally described the components and steps of each example according to their functions. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professionals and technicians may use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.

[0097] The above describes in detail a network interface card management system and server provided by the present application. This document uses specific examples to illustrate the principles and implementation methods of the present application. The description of the above embodiments is intended only to help understand the method and core concept of the present application. It should be noted that, for those skilled in the art, various improvements and modifications may be made to the present application without departing from the principles of the present application, and such improvements and modifications also fall within the scope of protection of the claims of the present application.

Claims

1. A network interface card management system, characterized in that: The system includes: a management module and a plurality of server computing nodes, each of the server computing nodes is connected to the management module, and the management module is connected to at least one of the network interface cards, wherein: The management module is configured to supply power to at least one of the network interface cards when the server is powered on; The plurality of server computing nodes are configured to send management signals of corresponding server computing nodes to the management module; The management module is further configured to receive the management signal and, if determining that at least one of the network interface cards is in the target operation mode, switch the connection state between each of the server computing nodes and the at least one network interface card according to the management signal; The plurality of server computing nodes include a plurality of first server computing nodes and a plurality of second server computing nodes, wherein the first server computing nodes represent server computing nodes in which the connection state between the node control unit and the network interface card is in a conducting state, and the second server computing nodes represent server computing nodes in which the connection state between the node control unit and the same network interface card is in a disconnected state. The management module includes an interface switching unit and a management unit, wherein the first node control unit of the first server computing node and the second node control unit of the second server computing node are respectively connected to the interface switching unit, the interface switching unit is connected to at least one of the network interface cards, and the management unit is connected to the interface switching unit, wherein: The second node control unit is configured to send an interface management signal to the management unit; The first node control unit is configured to receive the interface management signal sent by the management unit and send a response to the interface management signal to the management unit; The management unit is configured to, upon determining that at least one of the network interface cards is in the target operating mode, receive an interface management signal sent by the second node control unit, send the interface management signal to the first node control unit, and receive a response to the interface management signal sent by the first node control unit; The management unit is further configured to control the interface switching unit to switch the connection state between the second node control unit and at least one of the network interface cards according to a response to the interface management signal.

2. The network interface card management system according to claim 1, wherein: Each of the server computing nodes includes a corresponding node control unit and a central processing unit, and the management module includes a management unit, a power supply interface and a bus switching unit. The power supply interface is connected to at least one of the network interface cards, each of the node control units is connected to the management unit, the management unit is connected to at least one of the network interface cards, each of the central processing units is connected to the bus switching unit, the bus switching unit is connected to at least one of the network interface cards, and the management unit is connected to the bus switching unit. The management module is configured to control the power supply interface to supply power to at least one of the network interface cards when the server is powered on; The plurality of node control units are configured to collect power management signals of corresponding server computing nodes and send the corresponding power management signals to the management unit; The management unit is configured to receive the power management signal and, upon determining that at least one of the network interface cards is in a target operating mode, control the bus switching unit to switch the connection state between each of the central processing units and at least one of the network interface cards according to each of the power management signals.

3. The network interface card management system according to claim 2, wherein: The management unit is configured to control the bus switching unit to switch the connection state between the central processing unit and at least one of the network interface cards to a conducting state when the power management signal received is a first power management signal.

4. The network interface card management system according to claim 2, wherein: The management unit is configured to control the bus switching unit to switch the connection state between the central processing unit and at least one of the network interface cards to a disconnected state when the power management signal received is a second power management signal.

5. The network interface card management system according to claim 1, wherein: The management unit is configured to control the interface switching unit to switch the connection state between the second node control unit and at least one of the network interface cards to a conducting state when the response to the interface management signal received is a first response.

6. The network interface card management system according to claim 1, wherein: The management unit is configured to control the interface switching unit to switch the connection state between the second node control unit and at least one of the network interface cards to a disconnected state when the response to the interface management signal received is a second response.

7. The network interface card management system according to claim 1, wherein: The first node control unit is configured to send a first response to the management unit when the first node control unit does not require a network controller sideband interface service.

8. The network interface card management system according to claim 1, wherein: The first node control unit is configured to send a second response to the management unit when the first node control unit requires a network controller sideband interface service.

9. The network interface card management system according to claim 1 or 2, characterized in that: The management unit is configured to receive a power management signal from each of the server computing nodes and determine an operating mode of at least one of the network interface cards according to the power management signal.

10. The network interface card management system according to claim 9, characterized in that: The management unit is configured to determine that the operation mode is a target operation mode when a power management signal received from at least one of the server computing nodes is a target power management signal.

11. The network interface card management system according to claim 1, wherein: At least one of the network interface cards includes a storage unit, the management module includes a management unit, and the management unit is connected to the storage unit, wherein, The management unit is used to read the field replaceable unit information in the storage unit after the server is powered on, and control the server according to the field replaceable unit information.

12. The network interface card management system according to claim 11, characterized in that: The field replaceable unit information includes a mapping relationship between the temperature of the network interface card and the speed of the server fan, wherein: The management unit is configured to determine a target rotation speed of the server fan according to the temperature of the network interface card and a mapping relationship between the temperature of the network interface card and the rotation speed of the server fan, and control the server fan to operate at the target rotation speed.

13. The network interface card management system according to claim 11, characterized in that: The field replaceable unit information also includes a peripheral component interconnection express bus topology connection relationship of the network interface card, and the peripheral component interconnection express bus topology connection relationship of the network interface card represents the number of server computing nodes supported by the network interface card, wherein The management unit is used to control the network interface card to establish a connection with the server computing node according to the peripheral component interconnection fast bus topology connection relationship of the network interface card.

14. A server, characterized in that: A management system comprising the network interface card according to any one of claims 1 to 13.

Images