Computing power low-delay network transmission method of intelligent computing center and related equipment

Abstract

The invention provides a computing power low-delay network transmission method of an intelligent computing center and related equipment, and relates to the technical field of computing power of the intelligent computing center, and the intelligent computing center is configured with at least one remote direct memory access ROCE network card fused with the Ethernet. The method comprises the steps that S1, a configuration strategy corresponding to a target ROCE network card associated with a target object is obtained, the target object is a container or a virtual machine of the intelligent computing center, the target object is used for bearing a computing power task, the target ROCE network card is a ROCE network card in at least one ROCE network card, and the target ROCE network card is used for bearing the computing power task; the configuration strategy comprises a network connection mode of the target ROCE network card, and the network connection mode is an infinite bandwidth IB mode or an ROCE mode; and S2, allocating an interface resource of the target ROCE network card to the target object, wherein the interface resource is used for the target object to access the network through the target ROCE network card based on the network connection mode. According to the invention, the network transmission delay is reduced.

Description

Technical Field

[0001] The present invention relates to the field of computing power technology of an intelligent computing center, and in particular to a low-latency network transmission method and related equipment for computing power of an intelligent computing center. Background Art

[0002] "Intelligent computing center" refers to a facility that provides the required computing power, data and algorithms for artificial intelligence applications (such as artificial intelligence deep learning model development, model training and model reasoning) by using large-scale heterogeneous computing resources, including general computing power and intelligent computing power. The intelligent computing center covers facilities, hardware, and software, and can provide full-stack capabilities from bottom-level computing power to top-level application enablement.

[0003] In the intelligent computing center, containers (pods) or virtual machines (VMs) are set up to carry different computing tasks and implement the execution of different computing tasks. For example, containers or virtual machines carry different model training tasks and complete the training of different models.

[0004] At present, the data of containers or virtual machines in the intelligent computing center is connected to the network through the Ethernet (ETH) of the intelligent computing center to achieve data interaction. However, the data transmission of Ethernet can only adapt to application scenarios that are not sensitive to latency, and cannot meet application scenarios with high latency requirements.

[0005] It can be seen that the existing technology has the problem of large network transmission delay in the intelligent computing center. Summary of the invention

[0006] The embodiment of the present invention provides a low-latency network transmission method and related equipment for computing power of an intelligent computing center to solve the problem of large network transmission delay of the intelligent computing center in the prior art.

[0007] To solve the above problems, the present invention is achieved as follows:

[0008] In a first aspect, an embodiment of the present invention provides a method for low-latency network transmission of computing power in an intelligent computing center, wherein the intelligent computing center is configured with at least one converged Ethernet remote direct memory access (ROCE) network card, and the method comprises:

[0009] Step S1, obtaining a configuration policy corresponding to a target ROCE network card associated with a target object, wherein the target object is a container or a virtual machine of the intelligent computing center, the target object is used to carry computing tasks, the target ROCE network card is a ROCE network card among the at least one ROCE network card, and the configuration policy includes a network connection mode of the target ROCE network card, and the network connection mode is an infinite bandwidth IB mode or a ROCE mode;

[0010] Step S2: Allocate interface resources of the target ROCE network card to the target object, where the interface resources are used for the target object to access the network through the target ROCE network card based on the network connection mode.

[0011] In one embodiment, the method further comprises:

[0012] Step S3: when the configuration policy corresponding to the target ROCE network card is updated, obtaining the updated configuration policy;

[0013] Step S4: when the network connection mode included in the updated configuration policy changes, restart the host corresponding to the target ROCE network card to update the network connection mode of the target ROCE network card;

[0014] Wherein, after restarting the host corresponding to the target ROCE network card, the interface resource is used for the target object to access the network through the target ROCE network card based on the updated network connection mode.

[0015] In one embodiment, step S2 includes:

[0016] Step S21, obtaining configuration information corresponding to the target object;

[0017] Step S22: Allocate the interface resources of the target ROCE network card to the target object based on the configuration information.

[0018] In one embodiment, when the target object is the container, the configuration information includes a network namespace corresponding to the container; when the target object is the virtual machine, the configuration information includes a number corresponding to the interface resource of the target ROCE network card;

[0019] The step S22 comprises:

[0020] Step S221: when the target object is the container, allocate the interface resource to the target object, and switch the namespace of the interface resource to the network namespace corresponding to the container;

[0021] Step S222: when the target object is the virtual machine, allocate the interface resource to the target object based on the number.

[0022] In one embodiment, the target ROCE network card is associated with at least one target object, wherein:

[0023] In the case where the target ROCE network card is associated with a target object, the target object is directly connected to the network through the physical function PF;

[0024] In the case where the target ROCE network card is associated with multiple target objects, the multiple target objects are directly connected to the network through the virtual function VF.

[0025] In one embodiment, the target object corresponds to one configuration policy, and the configuration policy corresponds to one or more target objects.

[0026] In a second aspect, an embodiment of the present invention further provides a computing power low-latency network transmission device for an intelligent computing center, wherein the intelligent computing center is configured with at least one converged Ethernet remote direct memory access ROCE network card, and the device comprises:

[0027] A first acquisition module is used to acquire a configuration policy corresponding to a target ROCE network card associated with a target object, wherein the target object is a container or a virtual machine of the intelligent computing center, the target object is used to carry computing tasks, the target ROCE network card is a ROCE network card among the at least one ROCE network card, and the configuration policy includes a network connection mode of the target ROCE network card, and the network connection mode is an infinite bandwidth IB mode or an Ethernet ETH mode;

[0028] An allocation module is used to allocate interface resources of the target ROCE network card to the target object, and the interface resources are used for the target object to access the network through the target ROCE network card based on the network connection mode.

[0029] In a third aspect, an embodiment of the present invention further provides an electronic device comprising a processor, a memory, and a computer program stored in the memory and executable on the processor, wherein when the computer program is executed by the processor, the steps in the method for low-latency network transmission of computing power of an intelligent computing center as described in the first aspect above are implemented.

[0030] In a fourth aspect, an embodiment of the present invention further provides a computer-readable storage medium having a computer program stored thereon, and when the computer program is executed by a processor, the steps in the method for low-latency network transmission of computing power of an intelligent computing center as described in the first aspect above are implemented.

[0031] In a fifth aspect, an embodiment of the present invention further provides a computer program product, comprising computer instructions, which, when executed by a processor, implement the steps in the method for low-latency network transmission of computing power of an intelligent computing center as described in the first aspect above.

[0032] In an embodiment of the present invention, by obtaining a configuration policy corresponding to a target ROCE network card associated with a target object, the target object is a container or a virtual machine of an intelligent computing center, the target object is used to carry computing tasks, the target ROCE network card is a ROCE network card in at least one ROCE network card, the configuration policy includes a network connection mode of the target ROCE network card, the network connection mode is an infinite bandwidth IB mode or a ROCE mode, and an interface resource of the target ROCE network card is allocated to the target object. The interface resource is used for the target object to access the network through the target ROCE network card based on the network connection mode, so that the target object can access the network through the IB mode or the ROCE mode, so that the network can be accessed through the IB mode when there is a lot of data to be interacted, and the network can be accessed through the ROCE mode when there is less data to be interacted, thereby realizing rapid data interaction, thereby reducing the network transmission delay of the intelligent computing center. BRIEF DESCRIPTION OF THE DRAWINGS

[0033] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings required for use in the description of the embodiments of the present invention will be briefly introduced below. Obviously, the accompanying drawings in the following description are only some embodiments of the present invention. For ordinary technicians in this field, other accompanying drawings can be obtained based on these accompanying drawings without paying creative labor.

[0034] Figure 1 It is a flow chart of a method for low-latency network transmission of computing power of an intelligent computing center provided by an embodiment of the present invention;

[0035] Figure 2 It is a schematic diagram of computing task data interaction through an IB network provided by an embodiment of the present invention;

[0036] Figure 3 It is a structural diagram of a computing power low-latency network transmission device for an intelligent computing center provided by an embodiment of the present invention;

[0037] Figure 4 It is a structural diagram of an electronic device provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0038] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0039] The "computing power" mentioned in the present invention is the ability of computer equipment or computing / data center to process information. It is the ability of computer hardware and software to work together to execute certain computing requirements. It is the computing power to achieve target result output by processing information data. It is a new type of productivity that integrates information computing power, network carrying capacity, and data storage capacity. It mainly provides services to the society through computing power infrastructure.

[0040] The "computing power" (Computational Power, CP) described in the present invention is the ability of a data center server to process data and output results. It is a comprehensive indicator to measure the computing power of a data center, including general computing power, super computing power and intelligent computing power. The commonly used unit of measurement is the number of floating point operations performed per second (FLOPS: Floating Point Operations Per Second, 1EFLOPS = 10^18FLOPS). The larger the value, the stronger the comprehensive computing power. According to calculations, 1EFLOPS is approximately the computing power output of 5 Tianhe-2A or 500,000 mainstream server CPUs or 2 million mainstream notebooks. The calculation formula is: CP = CP 通用 +CP 智能 +CP 超级 .

[0041] The "Network Power" (NP) described in the present invention is a manifestation of the data transmission capability of computing facilities, including comprehensive capabilities such as network architecture, network bandwidth, transmission latency, intelligent management and scheduling, etc. The carrying capacity involves network transmission within and between data centers, and is a comprehensive indicator for measuring network transmission scheduling capabilities. In the embodiment of the present invention, the carrying capacity uses the memory bandwidth.

[0042] The "Storage Power" (SP) described in the present invention is the comprehensive ability of a data center in terms of data storage capacity, performance, safety and reliability, and green and low-carbon. It is a comprehensive indicator for measuring the data storage capacity of a data center, including external storage devices such as storage arrays and server built-in storage devices. The commonly used unit of measurement for storage capacity is exabyte (EB, 1EB = 2^60bytes), and the commonly used unit of measurement for performance is the number of reads and writes per second per unit capacity (IOPS / TB, Input / Output Operations Per Second / TB). The disaster recovery ratio is an important manifestation of safety and reliability.

[0043] The "computing power infrastructure" described in the present invention is a new type of information infrastructure that integrates information computing power, network carrying capacity, and data storage capacity. It can realize centralized computing, storage, transmission and application of information, and presents characteristics such as diversity and ubiquity, intelligence and agility, safety and reliability, green and low-carbon. It is of great significance to promote industrial transformation and upgrading, enable my country's scientific and technological innovation, meet people's better life and realize high-efficiency social governance.

[0044] The “computing power” mentioned in the present invention includes general computing power, intelligent computing power and super computing power.

[0045] The “general computing power” mentioned in the present invention refers to the computing power provided by a server based on a central processing unit (CPU) chip, which is used to support basic general computing such as cloud computing and edge computing.

[0046] The "intelligent computing power" described in the present invention is a computing platform for large-scale deployment of special chips such as graphics processing units (GPUs), field programmable gate arrays (FPGAs), and application-specific integrated circuits (ASICs) for various innovative applications of artificial intelligence, such as natural language processing and machine vision.

[0047] The "super computing power" mentioned in the present invention is mainly the computing power provided by high-performance computing clusters such as supercomputers. It utilizes the centralized computing resources of multiple computer systems working in parallel and uses a dedicated operating system to handle extremely complex or data-intensive problems. It is mainly used for calculations in cutting-edge scientific fields, such as planetary simulation, drug molecule design, and gene analysis.

[0048] The "intelligent computing center" mentioned in the present invention refers to a facility that provides the required computing power, data and algorithms for artificial intelligence applications (such as artificial intelligence deep learning model development, model training and model reasoning) by using large-scale heterogeneous computing resources, including general computing power (CPU) and intelligent computing power (GPU, FPGA, etc.). The intelligent computing center covers facilities, hardware, and software, and can provide full-stack capabilities from bottom-level computing power to top-level application enablement.

[0049] The "computing resources" mentioned in the present invention refer to the technologies and facilities with information calculation, transmission, storage and application capabilities required for the development of the digital society, including but not limited to computing resources such as CPUs and GPUs, network resources such as switches and routers, storage resources such as storage arrays and distributed storage, security resources such as firewalls and intrusion detection systems, and supporting and guarantee resources such as wind, fire, water and electricity.

[0050] The "low-latency network" mentioned in the present invention refers to networks such as InfiniBand (IB) and Remote Direct Memory Access over Converged Ethernet (RoCE). Low-latency networks can achieve ultra-low-latency data communication.

[0051] The "Remote Direct Memory Access over Converged Ethernet (ROCE) network card" described in the present invention is a network card for executing the ROCE protocol. The ROCE protocol is a cluster network communication protocol that can perform remote memory direct access (RDMA) on Ethernet, which can greatly reduce the latency of Ethernet communication and improve bandwidth utilization. As a core hardware device, ROCE is specially designed to efficiently support RDMA operations. By integrating necessary hardware features, it can free the CPU from heavy RDMA tasks, significantly reduce data transmission latency, and improve the operating performance of the entire system. The application scenarios of RoCE network cards include new business needs such as AI intelligent computing, high-performance computing, and high-speed storage.

[0052] The "pod" described in the present invention refers to a series of processes or tasks that are isolated from the rest of the system. All files required to run these processes or tasks are provided by another image, that is, the container is portable and consistent throughout the entire process from development to testing to production. Containers provide isolation between processes or tasks, and can divide the resources managed by computers, servers, or intelligent computing centers into isolated groups, resolving conflicts in resource usage between isolated groups.

[0053] The "virtual machine" described in the present invention is a computer system with complete hardware system functions simulated by software, which runs in a completely isolated environment. A virtual machine (VM) can be created on a physical computer, server or intelligent computing center, and a part of the hard disk and memory capacity of the physical computer, server or intelligent computing center is allocated to the virtual machine as its hard disk and memory capacity. Each virtual machine can perform the same operations as a physical computer. Virtual machine technology allows multiple operating systems to run simultaneously on a physical computer without interfering with each other.

[0054] The "physical function (PF)" described in the present invention is used to support the peripheral component interconnect (PCI) function in single root input / output virtualization (SR-IOV). As defined in the SR-IOV specification, PF includes a SR-IOV function configuration structure for managing the SR-IOV function. PF is a full-featured peripheral component interconnect Express (PCIe) function that can be discovered, managed, and processed like any other PCIe device. PF has fully configured resources and can be used to configure or control PCIe devices. The PF passthrough in the present invention is a physical function transmission channel corresponding to the physical interface of the network interface card, and a PF channel can correspond to a container or a virtual machine.

[0055] The "Virtual Function (VF)" described in the present invention is a lightweight PCIe function associated with a PF, which can share one or more physical resources with the PF and other VFs associated with the same PF. VF is only allowed to have configuration resources for its own behavior. In the present invention, the PF passthrough is associated with one or more VF passthroughs, and different virtual machines or containers correspond to a VF passthrough respectively, and the network is connected through the PF passthrough associated with the VF passthrough.

[0056] See also Figure 1 , Figure 1 is a flow chart of a method for low-latency network transmission of computing power in an intelligent computing center provided by an embodiment of the present invention. Figure 1 As shown, the intelligent computing center is configured with at least one converged Ethernet remote direct memory access ROCE network card, and the method includes the following steps:

[0057] Step S1, obtaining a configuration policy corresponding to a target ROCE network card associated with a target object, wherein the target object is a container or a virtual machine of the intelligent computing center, the target object is used to carry computing tasks, the target ROCE network card is a ROCE network card in the at least one ROCE network card, and the configuration policy includes a network connection mode of the target ROCE network card, and the network connection mode is an infinite bandwidth (InfiniBand, IB) mode or a ROCE mode;

[0058] Step S2: Allocate interface resources of the target ROCE network card to the target object, where the interface resources are used for the target object to access the network through the target ROCE network card based on the network connection mode.

[0059] In an embodiment of the present invention, by obtaining a configuration policy corresponding to a target ROCE network card associated with a target object, the target object is a container or a virtual machine of an intelligent computing center, the target object is used to carry computing tasks, the target ROCE network card is a ROCE network card in at least one ROCE network card, the configuration policy includes a network connection mode of the target ROCE network card, the network connection mode is an infinite bandwidth IB mode or a ROCE mode, and an interface resource of the target ROCE network card is allocated to the target object. The interface resource is used for the target object to access the network through the target ROCE network card based on the network connection mode, so that the target object can access the network through the IB mode or the ROCE mode, so that the network can be accessed through the IB mode when there is a lot of data to be interacted, and the network can be accessed through the ROCE mode when there is less data to be interacted, thereby realizing rapid data interaction, thereby reducing the network transmission delay of the intelligent computing center.

[0060] In an embodiment of the present invention, the IB mode of the ROCE network card can enable the target object to access the IB network through the interface resources of the target ROCE network card, and the ROCE mode of the ROCE network card can enable the target object to access the network corresponding to ROCE through the interface resources of the target ROCE network card. Among them, the data transmission speed of the IB network is faster than the data transmission speed of the network corresponding to ROCE, and the data transmission speed of the network corresponding to ROCE is faster than the data transmission speed of the ETH network. The data size that the power task needs to interact can be estimated, and then the network access mode that needs to be used for data transmission is determined according to the data size.

[0061] For example, Figure 2 As shown, the computing power tasks carried by the user-state container of host 1 and the computing power tasks carried by the user-state virtual machine of host 2 need to interact with each other. There is a lot of data to be interacted. At this time, the container of host 1 needs to access the IB network through the kernel state through the IB mode, and the virtual machine of host 2 needs to access the IB network through the kernel state through the IB mode. In this way, the computing power tasks carried by the container of host 1 and the virtual machines of the computing power tasks carried by host 2 can achieve data interaction through the IB network.

[0062] In an embodiment of the present invention, the intelligent computing center is configured with at least one ROCE network card, each of the at least one ROCE network card can be associated with at least one container or virtual machine, and the associated at least one container or virtual machine accesses the network through the interface resources of the ROCE network card. It should be noted that the network connection mode of the at least one associated container or virtual machine accessing the network is the same, and the configuration is implemented through the configuration policy of the ROCE network card.

[0063] In the embodiment of the present invention, the configuration strategy is a data code for configuring the target ROCE network card, and the configuration of the network connection mode of the target ROCE network card can be realized through the configuration strategy.

[0064] For example, the code for configuring the strategy to access the network through IB mode can be as follows:

[0065] Type:IB

[0066] ROCE No:

[0067] -ens12

[0068] The above code indicates that the ROCE network card numbered ens12 is connected to the network through the IB mode. At this time, the containers or virtual machines associated with the ROCE network card are connected to the network through the IB mode.

[0069] In one embodiment, the method further comprises:

[0070] Step S3: when the configuration policy corresponding to the target ROCE network card is updated, obtaining the updated configuration policy;

[0071] Step S4: when the network connection mode included in the updated configuration policy changes, restart the host corresponding to the target ROCE network card to update the network connection mode of the target ROCE network card;

[0072] Wherein, after restarting the host corresponding to the target ROCE network card, the interface resource is used for the target object to access the network through the target ROCE network card based on the updated network connection mode.

[0073] In an embodiment of the present invention, when the configuration policy corresponding to the target ROCE network card is updated, the updated configuration policy is obtained. When the network connection mode included in the updated configuration policy changes, the host corresponding to the target ROCE network card is restarted to update the network connection mode of the target ROCE network card. In this way, the network connection mode of the target object accessing the network is updated by updating the configuration policy, thereby adapting to different computing power tasks, and further improving the flexibility of computing power task data interaction in the intelligent computing center.

[0074] It should be noted that when the network connection mode needs to be updated (including updating the IB mode to the ROCE mode, and updating the ROCE mode to the IB mode), the configuration policy needs to be updated. After updating the configuration policy, in order to apply the configuration policy, the host corresponding to the target ROCE network card needs to be restarted. After the host is restarted, the target ROCE network card updates the network connection mode, so that the target object can access the network through the updated network connection mode.

[0075] For example, the updated configuration strategy accesses the network through the ROCE mode, and the code can be as follows:

[0076] Type:ROCE

[0077] ROCE No:

[0078] -ens12

[0079] The above code indicates that the ROCE network card numbered ens12 is changed from the original ROCE mode to the IB mode to access the network. At this time, the containers or virtual machines associated with the ROCE network card are all connected to the network through the ROCE mode.

[0080] Furthermore, in the case where the host includes multiple ROCE network cards, when the network connection mode of the configuration policy of one ROCE network card among the multiple ROCE network cards is updated, the host needs to be restarted to implement the updated configuration policy of the ROCE network card.

[0081] In one embodiment, step S2 includes:

[0082] Step S21, obtaining configuration information corresponding to the target object;

[0083] Step S22: Allocate the interface resources of the target ROCE network card to the target object based on the configuration information.

[0084] In an embodiment of the present invention, by obtaining the configuration information corresponding to the target object, and then allocating the interface resources of the target ROCE network card to the target object based on the configuration information, the target object carrying the computing task can access the network through the interface resources of the target ROCE network card.

[0085] It should be noted that the interface resources of the target ROCE network card are the resources of the physical interface of the target ROCE network card. The number of interface resources of the target ROCE network card is determined according to the number of physical interfaces of the target ROCE network card and corresponds one to one. After the interface resources of the target ROCE network card are allocated to the target object, the target object accesses the network through the physical interface corresponding to the interface resources.

[0086] Among them, when the interface resources of the target ROCE network card are occupied by data transmission by other containers or virtual machines, the target object needs to wait for the data transmission to be completed before transmitting data through the interface resources of the target ROCE network card.

[0087] In one embodiment, when the target object is the container, the configuration information includes a network namespace corresponding to the container; when the target object is the virtual machine, the configuration information includes a number corresponding to the interface resource of the target ROCE network card;

[0088] The step S22 comprises:

[0089] Step S221: when the target object is the container, allocate the interface resource to the target object, and switch the namespace of the interface resource to the network namespace corresponding to the container;

[0090] Step S222: when the target object is the virtual machine, allocate the interface resource to the target object based on the number.

[0091] In an embodiment of the present invention, when the target object is a container, interface resources are allocated to the target object, and the namespace of the interface resources is switched to the network namespace corresponding to the container, so that the container can access the network through the interface resources of the target ROCE network card; when the target object is a virtual machine, interface resources are allocated to the target object based on the number, so that the virtual machine can access the network through the interface resources of the target ROCE network card.

[0092] It should be noted that the ROCE network card can be associated with different types of target objects. When different types of target objects need to access the network, according to the embodiment of the present invention, the network access is achieved through the interface resources of the ROCE network card in different time periods.

[0093] In one embodiment, the target ROCE network card is associated with at least one target object, wherein:

[0094] In the case where the target ROCE network card is associated with a target object, the target object is directly connected to the network through the physical function PF;

[0095] In the case where the target ROCE network card is associated with multiple target objects, the multiple target objects are directly connected to the network through the virtual function VF.

[0096] In an embodiment of the present invention, when the target ROCE network card is associated with one target object, the target object accesses the network through the physical function PF; when the target ROCE network card is associated with multiple target objects, the multiple target objects access the network through the virtual function VF, thereby realizing data interaction between computing tasks carried by different target objects through the network.

[0097] For example, when the ROCE network card of host 1 is associated with only one container, and the ROCE network card of host 2 is associated with only multiple virtual machines, the container bearer is directly connected to the network through PF, and the virtual machine is directly connected to the network through VF.

[0098] In one embodiment, the target object corresponds to one configuration policy, and the configuration policy corresponds to one or more target objects.

[0099] In an embodiment of the present invention, a target object corresponds to a configuration policy, and a configuration policy corresponds to one or more target objects. Thus, configuration of different ROCE network cards is implemented through one configuration policy. Meanwhile, different ROCE network cards correspond to different target objects, thereby avoiding mutual interference between multiple configuration policies. When it is necessary to change the network connection mode of the target object accessing the network, it is only necessary to obtain the configuration policy of the ROCE network card associated with the target object, and then update the configuration policy, thereby realizing simple and fast completion of the update of the configuration policy and the update of the network connection mode of the target object accessing the network.

[0100] See also Figure 3 , Figure 3 is a structural diagram of a computing power low-latency network transmission device for an intelligent computing center provided by an embodiment of the present invention, such as Figure 3 As shown, the intelligent computing center is configured with at least one converged Ethernet remote direct memory access ROCE network card, and the device 300 includes:

[0101] The first acquisition module 301 is used to acquire a configuration policy corresponding to a target ROCE network card associated with a target object, wherein the target object is a container or a virtual machine of the intelligent computing center, the target object is used to carry computing tasks, the target ROCE network card is a ROCE network card among the at least one ROCE network card, and the configuration policy includes a network connection mode of the target ROCE network card, and the network connection mode is an infinite bandwidth IB mode or a ROCE mode;

[0102] The allocation module 302 is used to allocate the interface resources of the target ROCE network card to the target object, and the interface resources are used for the target object to access the network through the target ROCE network card based on the network connection mode.

[0103] In one embodiment, the apparatus 300 further includes:

[0104] A second acquisition module is used to acquire the updated configuration policy when the configuration policy corresponding to the target ROCE network card is updated;

[0105] A restart module, configured to restart the host corresponding to the target ROCE network card when the network connection mode included in the updated configuration policy changes, so as to update the network connection mode of the target ROCE network card;

[0106] Wherein, after restarting the host corresponding to the target ROCE network card, the interface resource is used for the target object to access the network through the target ROCE network card based on the updated network connection mode.

[0107] In one embodiment, the allocation module 302 includes:

[0108] An acquisition unit, used to acquire configuration information corresponding to the target object;

[0109] An allocating unit is configured to allocate the interface resource of the target ROCE network card to the target object based on the configuration information.

[0110] In one embodiment, when the target object is the container, the configuration information includes a network namespace corresponding to the container; when the target object is the virtual machine, the configuration information includes a number corresponding to the interface resource of the target ROCE network card;

[0111] The distribution unit comprises:

[0112] A first allocation subunit is configured to allocate the interface resource to the target object when the target object is the container, and switch the namespace of the interface resource to a network namespace corresponding to the container;

[0113] The second allocation subunit is configured to allocate the interface resource to the target object based on the number when the target object is the virtual machine.

[0114] In one embodiment, the target ROCE network card is associated with at least one target object, wherein:

[0115] In the case where the target ROCE network card is associated with a target object, the target object is directly connected to the network through the physical function PF;

[0116] In the case where the target ROCE network card is associated with multiple target objects, the multiple target objects are directly connected to the network through the virtual function VF.

[0117] In one embodiment, the target object corresponds to one configuration policy, and the configuration policy corresponds to one or more target objects.

[0118] The low-latency network transmission device for computing power of an intelligent computing center provided in an embodiment of the present invention is capable of implementing the various processes of each embodiment of the low-latency network transmission method for computing power of the above-mentioned intelligent computing center. The technical features correspond one to one and can achieve the same technical effect. To avoid repetition, they will not be described here.

[0119] It should be noted that the computing power low-latency network transmission device of the intelligent computing center in the embodiment of the present invention can be a device, or it can be a component, integrated circuit, or chip in an electronic device.

[0120] The embodiment of the present invention further provides an electronic device, see Figure 4 , Figure 4 4 is a schematic diagram of a structure of an electronic device provided by the present invention, the electronic device includes a memory 401, a processor 402 and a program or instruction stored in the memory 401 and running thereon, and when the program or instruction is executed by the processor 402, it can be realized Figure 1 Any steps in the corresponding intelligent computing center's computing power low-latency network transmission embodiment and the same beneficial effects achieved will not be repeated here.

[0121] The processor 502 may be a CPU, an ASIC, an FPGA or a GPU.

[0122] A person skilled in the art can understand that all or part of the steps of the embodiment of the method for low-latency network transmission of computing power of the intelligent computing center can be completed by hardware related to program instructions, and the program can be stored in a readable medium.

[0123] The embodiment of the present invention further provides a computer-readable storage medium, wherein a computer program is stored on the computer-readable storage medium, and when the computer program is executed by a processor, the above-mentioned Figure 1 Any step in the corresponding embodiment of the low-latency network transmission method of the computing power of the intelligent computing center can achieve the same technical effect. To avoid repetition, it is not repeated here. The storage medium is such as a read-only memory (ROM), a random access memory (RAM), a disk or an optical disk.

[0124] The embodiment of the present invention also provides a computer program product, including computer instructions, which can achieve the above-mentioned Figure 1 Any step in the corresponding embodiment of the low-latency network transmission method of the computing power of the intelligent computing center can achieve the same technical effect. To avoid repetition, it will not be repeated here.

[0125] The terms "first", "second" etc. in the embodiments of the present invention are used to distinguish similar objects, and need not be used to describe a specific order or sequential order. In addition, the terms "include" and "have" and any variation thereof are intended to cover non-exclusive inclusions, for example, the process, method, system, product or equipment comprising a series of steps or units need not be limited to those steps or units clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or equipment. In addition, "and / or" is used in the present application to represent at least one of the connected objects, such as A and / or B and / or C, indicating that A alone, B alone, C alone, and A and B all exist, B and C all exist, A and C all exist, and 7 situations in which A, B and C all exist.

[0126] It should be noted that, in this article, the terms "include", "comprises" or any other variations thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device. In the absence of further restrictions, an element defined by the sentence "comprises a ..." does not exclude the existence of other identical elements in the process, method, article or device including the element.

[0127] Through the description of the above implementation methods, those skilled in the art can clearly understand that the above-mentioned embodiment methods can be implemented by means of software plus a necessary general hardware platform, and of course by hardware, but in many cases the former is a better implementation method. Based on such an understanding, the technical solution of the present application, or the part that contributes to the prior art, can be embodied in the form of a software product, which is stored in a storage medium (such as ROM / RAM, a disk, or an optical disk), and includes a number of instructions for a terminal (which can be a mobile phone, a computer, a server, an air conditioner, or a second terminal device, etc.) to execute the methods of each embodiment of the present application.

[0128] The embodiments of the present application are described above in conjunction with the accompanying drawings, but the present application is not limited to the above-mentioned specific implementation methods. The above-mentioned specific implementation methods are merely illustrative and not restrictive. Under the guidance of the present application, ordinary technicians in this field can also make many forms without departing from the purpose of the present application and the scope of protection of the claims, all of which are within the protection of the present application.

Claims

1. A method for low-latency network transmission of computing power in an intelligent computing center, wherein the intelligent computing center is configured with at least one converged Ethernet remote direct memory access (ROCE) network card, characterized in that: The method comprises: Step S1, obtaining a configuration policy corresponding to a target ROCE network card associated with a target object, wherein the target object is a container or a virtual machine of the intelligent computing center, the target object is used to carry computing tasks, the target ROCE network card is a ROCE network card among the at least one ROCE network card, and the configuration policy includes a network connection mode of the target ROCE network card, and the network connection mode is an infinite bandwidth IB mode or a ROCE mode; Step S2: Allocate interface resources of the target ROCE network card to the target object, where the interface resources are used for the target object to access the network through the target ROCE network card based on the network connection mode.

2. The method according to claim 1, characterized in that The method further comprises: Step S3: when the configuration policy corresponding to the target ROCE network card is updated, obtaining the updated configuration policy; Step S4: when the network connection mode included in the updated configuration policy changes, restart the host corresponding to the target ROCE network card to update the network connection mode of the target ROCE network card; Wherein, after restarting the host corresponding to the target ROCE network card, the interface resource is used for the target object to access the network through the target ROCE network card based on the updated network connection mode.

3. The method according to claim 1 or 2, characterized in that The step S2 comprises: Step S21, obtaining configuration information corresponding to the target object; Step S22: Allocate the interface resources of the target ROCE network card to the target object based on the configuration information.

4. The method according to claim 3, characterized in that When the target object is the container, the configuration information includes a network namespace corresponding to the container; when the target object is the virtual machine, the configuration information includes a number corresponding to the interface resource of the target ROCE network card; The step S22 comprises: Step S221: when the target object is the container, allocate the interface resource to the target object, and switch the namespace of the interface resource to the network namespace corresponding to the container; Step S222: when the target object is the virtual machine, allocate the interface resource to the target object based on the number.

5. The method according to claim 3, characterized in that The target ROCE network card is associated with at least one target object, wherein: In the case where the target ROCE network card is associated with a target object, the target object is directly connected to the network through the physical function PF; In the case where the target ROCE network card is associated with multiple target objects, the multiple target objects are directly connected to the network through the virtual function VF.

6. The method according to claim 1 or 2, characterized in that: The target object corresponds to one configuration policy, and the configuration policy corresponds to one or more target objects.

7. A computing power low-latency network transmission device for an intelligent computing center, wherein the intelligent computing center is configured with at least one converged Ethernet remote direct memory access (ROCE) network card, characterized in that: The device comprises: A first acquisition module is used to acquire a configuration policy corresponding to a target ROCE network card associated with a target object, wherein the target object is a container or a virtual machine of the intelligent computing center, the target object is used to carry computing tasks, the target ROCE network card is a ROCE network card among the at least one ROCE network card, and the configuration policy includes a network connection mode of the target ROCE network card, and the network connection mode is an infinite bandwidth IB mode or a ROCE mode; An allocation module is used to allocate interface resources of the target ROCE network card to the target object, and the interface resources are used for the target object to access the network through the target ROCE network card based on the network connection mode.

8. An electronic device, characterized in that: include: A processor, a memory, and a program stored in the memory and executable on the processor, wherein when the program is executed by the processor, the steps of the method for low-latency network transmission of computing power of an intelligent computing center as described in any one of claims 1 to 7 are implemented.

9. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a computer program, which, when executed by a processor, implements the steps of the method for low-latency network transmission of computing power in an intelligent computing center according to any one of claims 1 to 6.

10. A computer program product, characterized in that It includes computer instructions, which, when executed by a processor, implement the steps of the low-latency network transmission method for computing power of an intelligent computing center as described in any one of claims 1 to 6.

Images