A job processing method and related apparatus

Abstract

Embodiments of the present application disclose a job processing method and related equipment, which are used for configuring a cloud computing instance with RoCE function for a local network to improve the efficiency of high-performance computing. In the job processing method, a cloud platform creates a cloud computing instance with RoCE function according to a received resource application, and sets the cloud computing instance to access the local network, so that the cloud computing instance processes data corresponding to a task in the local network. Since the cloud computing instance can receive the task from the local network and access the data corresponding to the task in the local network through the operating system, the cloud platform does not need to establish an HPC cluster on the cloud, and the local network does not need to transmit the task and the data corresponding to the task to the cloud platform. Therefore, the efficiency of high-performance computing is improved.

Description

TECHNICAL FIELD

[0001] Embodiments of the present application relate to the field of cloud computing, and in particular, to a job processing method and related equipment. BACKGROUND

[0002] Remote direct memory access (RDMA) is a technology for accessing data in the memory of a remote host by bypassing the kernel of the operating system of the remote host. The RDMA technology can save processing resources, improve system throughput, and reduce network communication delay of the system. The RDMA technology has multiple implementation manners, one of which is remote direct memory access over converged ethernet (RDMA over converged ethernet, RoCE). The RoCE technology is often applied in high-performance computing (HPC) based on a unified cluster.

[0003] In the prior art, when the computing resources of a local HPC cluster based on a RoCE network are insufficient to support high-performance computing, the local HPC cluster will apply for computing resources from a cloud platform. Then, the cloud platform will rebuild an HPC cluster on the cloud, and the cloud platform will send tasks from the local HPC cluster and data corresponding to the tasks to a cloud resource control node in the HPC cluster on the cloud, so that the cloud resource control node allocates the tasks and the data corresponding to the tasks to cloud computing nodes, so that the cloud computing nodes perform high-performance computing according to the tasks.

[0004] However, since the HPC cluster on the cloud and the local HPC cluster are two independent clusters, the local HPC cluster needs to transmit the tasks and the data corresponding to the tasks to the HPC cluster on the cloud, so that the HPC cluster on the cloud can perform high-performance computing. In this process, a local resource control node in the local HPC cluster needs to divide the local tasks. However, the tasks in the process of high-performance computing are heavy and difficult to divide, and the transmission of the data corresponding to the tasks to the cloud platform will also be limited by the transmission bandwidth. Therefore, the foregoing scheme cannot effectively improve the efficiency of high-performance computing. SUMMARY

[0005] Embodiments of the present application provide a job processing method and related equipment for configuring a cloud computing instance with RoCE function for a local network, so as to improve the efficiency of high-performance computing.

[0006] In a first aspect, an embodiment of the present application provides a job processing method, which can be applied to a high-performance computing scenario. In the job processing method, a cloud platform receives a resource application under a preset condition. The resource application is used to indicate that the cloud platform creates a cloud computing instance with a remote direct memory access over converged Ethernet (RoCE) function. Then, the cloud platform creates the cloud computing instance with the RoCE function according to the resource application, and sets the cloud computing instance to access a local network, so that the cloud computing instance processes data corresponding to a task in the local network.

[0007] The preset condition can include one or more of the following:

[0008] The computing resources of the local network are insufficient to support high-performance computing, or a user submits a resource application to the cloud platform according to the user's own needs, and the like. The specific conditions are not limited here.

[0009] In the embodiment of the present application, the cloud platform can create the cloud computing instance with the RoCE function according to the resource application, and access the local network with the cloud computing instance. Therefore, the cloud computing instance can receive the task from the local network beyond the operating system, and access the data corresponding to the task in the local network beyond the operating system, to process the data corresponding to the task in the local network. Therefore, the cloud platform does not need to establish an HPC cluster on the cloud, and the local network does not need to transmit the task and the data corresponding to the task to the cloud platform. Therefore, it is beneficial to improve the efficiency of high-performance computing.

[0010] Based on the first aspect, in an optional implementation, the local network is provided with a task issuing node and a task data storage node. At this time, in the job processing method, the cloud computing instance can receive the task sent by the task issuing node; then, the cloud computing instance acquires the task data corresponding to the task from the task data storage node through remote direct memory access (RDMA) according to the task, and performs task processing.

[0011] In the embodiment, two nodes with different functions in the local network are proposed, wherein the task issuing node is configured to issue tasks to the local computing nodes; and the task data storage node is configured to store data corresponding to the tasks, and the local computing nodes can obtain the data corresponding to the tasks from the task data storage node. Since the cloud computing instance has accessed the local network, the cloud computing instance is essentially connected to the task issuing node and the task data storage node. Therefore, the task issuing node can issue tasks to the cloud computing instance. Since the cloud computing instance has the RoCE function, the cloud computing instance can access the data corresponding to the tasks in the task data storage node without the operating system, and the cloud computing instance can process the data corresponding to the tasks. In such an embodiment, the cloud computing instance is actually added to the local HPC cluster as a computing node. Since the cloud computing instance and the local computing nodes form an HPC cluster, the task issuing node can assign tasks to the cloud computing instance as a local computing node. Therefore, the task issuing node does not need to transmit tasks and data corresponding to the tasks to the cloud platform. Therefore, the efficiency of high-performance computing is improved.

[0012] Based on the first aspect or the optional embodiment, in another optional embodiment, before the cloud platform receives the resource application, the method further includes: the task issuing node confirming that the number of tasks to be processed exceeds a threshold.

[0013] In the embodiment, the task issuing node can monitor the number of tasks to be processed, and measure the number of tasks to be processed and the computing capacity of the local computing nodes. When the task issuing node confirms that the number of tasks to be processed exceeds a threshold, that is, when the computing capacity (or computing resource) of the local computing nodes is insufficient to support the tasks to be processed, the task issuing node triggers the step of sending a resource application to the cloud platform.

[0014] It should be understood that the tasks to be processed can be tasks that have not been assigned to the local computing nodes by the task issuing node, or can be total tasks (including tasks that have been assigned to the local computing nodes) that need to be processed within a certain time range, which is not limited here. Of course, when the tasks represent different meanings, the threshold value will also be different, and the task issuing node can be adjusted according to the actual needs, which is not limited here.

[0015] Based on the first aspect or the optional embodiment, in another optional embodiment, the method further includes: the cloud platform receiving a resource revocation request sent by the task issuing node, wherein the resource revocation request is configured to indicate that the task has been completed or the task does not need to be executed. Then, the cloud platform revokes the cloud computing instance according to the resource revocation request.

[0016] In the embodiment, when the task is completed or the task does not need to be executed, the task issuing node can send a resource revocation request to the cloud platform, so that the cloud platform revokes the cloud computing instance according to the revocation request. In this scheme, the task issuing node can revoke the cloud computing instance, and the cloud platform can also allocate the cloud computing instance to other clusters. Therefore, the cloud computing instance configured by the cloud platform can be more flexible, and the utilization rate of the cloud computing instance can be improved.

[0017] Based on the first aspect or the optional implementation, in another optional implementation, the cloud platform creates the cloud computing instance with RoCE function according to the resource application, including: the cloud platform obtains a RoCE software package. Then, the cloud platform sends the RoCE software package to the initial cloud computing instance, and triggers the RoCE software package to be installed in the initial cloud computing instance to obtain the cloud computing instance with RoCE function.

[0018] The RoCE software package can be pre-stored in the cloud platform or pre-stored in the initial cloud computing instance, which is not limited here.

[0019] In the embodiment, a specific implementation of configuring the initial cloud computing instance without RoCE network card with RoCE function is proposed. Specifically, the cloud platform can install a RoCE software package in the initial cloud computing instance, so that the initial cloud computing instance simulates the RoCE network card by running the RoCE software. In this implementation, the cloud computing instance can use a common network card without using a RoCE network card. Therefore, the configuration cost of the cloud computing instance can be saved, and the realizability of the scheme can be improved.

[0020] Based on the first aspect or the optional implementation, in another optional implementation, the resource application includes configuration information, and the configuration information is used to indicate the configuration of the cloud computing instance required by the local network. The method further includes: the cloud platform creates an initial cloud computing instance according to the configuration information; or the cloud platform selects an initial cloud computing instance corresponding to the configuration information from a plurality of cloud computing instances according to the configuration information.

[0021] In the embodiment, the configuration information is included in the resource application. The cloud platform can create an initial cloud computing instance based on the configuration information, or can find an initial cloud computing instance matching the configuration information from the existing initial cloud computing instances based on the configuration information.

[0022] Based on the first aspect or the optional implementation, in another optional implementation, the cloud computing instance is a virtual machine, a container or a bare metal server.

[0023] In a second aspect, an embodiment of the present application provides a cloud platform, comprising: a processor, a network interface and a memory. The memory is configured to store data and program codes. The network interface is configured to receive a resource application, the resource application being configured to instruct the cloud platform to create a cloud computing instance with a remote direct memory access over converged Ethernet (RoCE) function. The processor is configured to create the cloud computing instance with the RoCE function according to the resource application, and set the cloud computing instance to access a local network.

[0024] In the embodiment of the present application, the cloud platform can create the cloud computing instance with the RoCE function according to the resource application, and set the cloud computing instance to access the local network. Therefore, the cloud computing instance can receive a task from the local network without going through the operating system, and access data corresponding to the task in the local network without going through the operating system, so as to process the data corresponding to the task in the local network. Therefore, the cloud platform does not need to establish an HPC cluster on the cloud, and the local network does not need to transmit the task and the data corresponding to the task to the cloud platform. Therefore, the efficiency of high-performance computing is improved.

[0025] Based on the second aspect, in an optional implementation, the local network is provided with a task distribution node and a task data storage node. The task distribution node is configured to send a task to the cloud computing instance. The task data storage node is configured to provide task data corresponding to the task to the cloud computing instance through remote direct memory access (RDMA).

[0026] Based on the second aspect or the optional implementation, in another optional implementation, the network interface is specifically configured to receive the resource application from the task distribution node when the task distribution node confirms that the number of tasks to be processed exceeds a threshold.

[0027] Based on the second aspect or the optional implementation, in another optional implementation, the network interface is further configured to receive a resource revocation request from the task distribution node, the resource revocation request being configured to indicate that the task has been completed or the task does not need to be executed. The processor is further configured to revoke the cloud computing instance according to the resource revocation request.

[0028] Based on the second aspect or the optional implementation, in another optional implementation, the processor is specifically configured to: obtain a RoCE software package; control the network interface to send the RoCE software package to an initial cloud computing instance, and trigger the RoCE software package to be installed in the initial cloud computing instance to obtain the cloud computing instance with the RoCE function.

[0029] In an alternative implementation based on the foregoing second aspect or the foregoing alternative implementation, the resource application comprises configuration information used to indicate a configuration of a cloud computing instance required by the local network. In addition, the processor is further configured to: create an initial cloud computing instance according to the configuration information; or select an initial cloud computing instance corresponding to the configuration information from a plurality of cloud computing instances according to the configuration information.

[0030] In an alternative implementation based on the foregoing second aspect or the foregoing alternative implementation, the cloud computing instance is a virtual machine, a container, or a bare-metal server.

[0031] It should be noted that the embodiments of the present application have a variety of other specific implementations, which can be specifically referred to the specific implementations and advantages of the first aspect, and will not be described here.

[0032] In a third aspect, the embodiments of the present application provide a cloud platform, which comprises:

[0033] a receiving module configured to receive a resource application, the resource application being used to indicate that the cloud platform creates a cloud computing instance with a Remote Direct Memory Access over Converged Ethernet (RoCE) function;

[0034] a resource configuration module configured to create the cloud computing instance with the RoCE function according to the resource application, and set the cloud computing instance to access the local network.

[0035] In the embodiments of the present application, since the cloud platform can create the cloud computing instance with the RoCE function according to the resource application, and access the local network, the cloud computing instance can receive tasks from the local network and access data corresponding to the tasks in the local network through the operating system, so as to process the data corresponding to the tasks in the local network. Therefore, the cloud platform does not need to establish an HPC cluster on the cloud, and the local network does not need to transmit tasks and data corresponding to the tasks to the cloud platform. Therefore, it is beneficial to improve the efficiency of high-performance computing.

[0036] In an alternative implementation based on the foregoing third aspect, the local network is provided with a task issuing node and a task data storage node. The task issuing node is configured to send tasks to the cloud computing instance, and the task data storage node is configured to provide task data corresponding to the tasks to the cloud computing instance through Remote Direct Memory Access (RDMA).

[0037] In an alternative implementation based on the foregoing third aspect or the foregoing alternative implementation, the receiving module is specifically configured to: when the task issuing node confirms that the number of tasks to be processed exceeds a threshold value, the transceiver receives the resource application from the task issuing node.

[0038] In an optional implementation based on the third aspect or the optional implementation, the receiving module is further configured to receive a resource revocation request from the task issuing node, the resource revocation request being used to indicate that the task has been completed or the task does not need to be executed; and the resource configuring module is further configured to revoke the cloud computing instance according to the resource revocation request.

[0039] In an optional implementation based on the third aspect or the optional implementation, the resource configuring module is specifically configured to: obtain a RoCE software package; control the transceiver to send the RoCE software package to an initial cloud computing instance, and trigger the RoCE software package to be installed in the initial cloud computing instance to obtain the cloud computing instance with the RoCE function.

[0040] In an optional implementation based on the third aspect or the optional implementation, the resource application includes configuration information, the configuration information being used to indicate a configuration of the cloud computing instance required by the local network. In addition, the resource configuring module is further configured to: create an initial cloud computing instance according to the configuration information; or find an initial cloud computing instance corresponding to the configuration information from a plurality of initial cloud computing instances according to the configuration information.

[0041] In an optional implementation based on the third aspect or the optional implementation, the cloud computing instance is a virtual machine, a container or a bare metal server.

[0042] It should be noted that the embodiments of the present application have a variety of other specific implementations, and specific implementations and advantages thereof can be referred to the specific implementations and advantages of the first aspect, which will not be described here.

[0043] In a fourth aspect, the embodiments of the present application provide a job processing system, the job system including a cloud platform, a cloud computing instance and a local network, wherein the local network includes a task issuing node and a task data storage node. In the job processing system: the cloud platform is configured to receive a resource application, create a cloud computing instance with a RoCE function according to the resource application, and set the cloud computing instance to access the local network, the resource application being used to indicate that the cloud platform creates a cloud computing instance with a RoCE function based on a converged Ethernet remote direct memory access. The cloud computing instance is configured to receive a task sent by the task issuing node, and obtain task data corresponding to the task from the task data storage node through a remote direct memory access RDMA according to the task, and perform task processing.

[0044] In addition, other functions of the cloud platform in the job processing system can refer to the embodiments of the first aspect or the embodiments of the second aspect; and other functions of the cloud computing instance in the job processing system can refer to the embodiments of the first aspect or the embodiments of the second aspect.

[0045] In a fifth aspect, the embodiments of the present application provide a bare metal server, which is a computing server with elasticity of virtual machines and performance of physical machines, and is used to provide excellent computing performance and data security for core databases, key application systems, high-performance computing, big data and other services. The bare metal server comprises a processing module and a transceiving module. The processing module can be a processor, and the transceiving module can be an input / output device or a network interface. The bare metal server can further comprise a storage module, which can be a memory. The storage module is used to store instructions, and the processing module executes the instructions stored in the storage module, so that the bare metal server performs the functions of the cloud computing instance in the first aspect or the second aspect.

[0046] In a sixth aspect, the embodiments of the present application provide a physical machine, which is used to create a virtual machine or a container based on the resource application in the first aspect or the second aspect. The physical machine comprises a processing module and a transceiving module. The processing module can be a processor, and the transceiving module can be an input / output device or a network interface. The physical machine can further comprise a storage module, which can be a memory. The storage module is used to store instructions, and the processing module executes the instructions stored in the storage module, so that the physical machine performs the functions of the cloud computing instance in the first aspect or the second aspect.

[0047] From the above technical solutions, the embodiments of the present application have the following advantages:

[0048] In the embodiments of the present application, the cloud platform can set a cloud computing instance with RoCE function according to the resource application, and access the cloud computing instance to the local network. Therefore, the cloud computing instance can receive tasks from the local network and access data corresponding to the tasks in the local network through the operating system, so as to process the data corresponding to the tasks in the local network. Therefore, the cloud platform does not need to establish an HPC cluster on the cloud, and the local network does not need to transmit tasks and data corresponding to the tasks to the cloud platform. Therefore, it is beneficial to improve the efficiency of high-performance computing. BRIEF DESCRIPTION OF DRAWINGS

[0049] In order to more clearly illustrate the technical solutions of the embodiments of the present application, the drawings needed in the embodiment description will be briefly introduced. Obviously, the drawings in the following description are only some embodiments of the present application.

[0050] Figure 1 A system architecture diagram of a job processing method in an embodiment of the present application;

[0051] Figure 2 A flowchart of a job processing method in an embodiment of the present application;

[0052] Figure 3A Another flowchart of a job processing method in an embodiment of the present application;

[0053] Figure 3B A logical connection diagram between a local HPC cluster and a cloud computing instance in an embodiment of the present application;

[0054] Figure 4 An embodiment diagram of a cloud platform in an embodiment of the present application;

[0055] Figure 5 Another embodiment diagram of a cloud platform in an embodiment of the present application. DETAILED DESCRIPTION

[0056] The technical solutions in the embodiments of the present application will be described clearly and completely below with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are only some of the embodiments of the present application, not all the embodiments.

[0057] The terms "first", "second", "third", "fourth" and the like (if any) in the specification and claims of the present application and the above drawings are used to distinguish similar objects, and do not necessarily indicate a specific order or a chronological sequence. It should be understood that the data thus used can be interchanged under appropriate circumstances, so that the embodiments described herein can be implemented in an order other than that illustrated or described herein. In addition, the terms "include" and "have" and any variations thereof are intended to cover non-exclusive inclusion, for example, a process, method, system, product or device that includes a series of steps or units does not necessarily have to be limited to those steps or units clearly listed, but can include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices.

[0058] The embodiments of the present application provide a job processing method and related equipment, which are used to configure a cloud computing instance with RoCE function for a local network, so as to improve the efficiency of high-performance computing.

[0059] For ease of understanding, the professional terms involved in the embodiments of the present application will be explained first as follows:

[0060] Remote Direct Memory Access (RDMA): a data transmission technology that can quickly move data from the memory of one machine or device to the memory of another machine or device, without transmitting data over the network through the operating system kernel protocol stack and without any impact on the operating system. Common RDMA implementation forms include virtual interface architecture, RDMA over converged Ethernet (RoCE), Infiniband (IB), and iWARP. Among them, the RoCE technology avoids the copying process of data between the user space and the kernel space and the processing process of data in the kernel protocol stack, can reduce memory consumption and CPU consumption, and also can reduce the delay of data transmission.

[0061] High Performance Computing (HPC): refers to using aggregated computing power to process data-intensive computing tasks that standard workstations cannot complete, including simulation, modeling, and rendering, etc. The high performance computing in the embodiments of the present application is RoCE-based high performance computing. In the embodiments of the present application, the cluster composed of devices implementing the foregoing high performance computing is referred to as an HPC cluster.

[0062] Cloud Platform: refers to an entity that provides services based on hardware resources and / or software resources to a remote device. The cloud platform in the embodiments of the present application can be a storage-type cloud platform mainly for data storage, a computing-type cloud platform mainly for data processing, or a comprehensive cloud computing platform that takes into account both computing and data storage processing.

[0063] Cloud Computing Instance: in the embodiments of the present application, refers to a computing resource created by a cloud platform to support task processing of a local network (for example, an HPC cluster).

[0064] The application scenarios and system architecture to which the job processing method proposed in the embodiments of the present application are suitable are introduced as follows:

[0065] The job processing method proposed in the embodiments of the present application is mainly applied to the scenario of an HPC cluster based on a RoCE network applying for computing resources from a cloud platform. For example, Figure 1As shown, the local network (i.e., the HPC cluster) in the scenario mainly includes a task dispatching node 101, a task data storage node 102, and a plurality of local task processing nodes 103. The task dispatching node 101 is configured to assign a computing task to each local task processing node 103 in the HPC cluster. Each local task processing node 103 in the local network is provided with a RoCE function, and thus, the aforementioned local task processing node 103 can access the task data in the task data storage node 102 over the operating system. In addition, the task dispatching node 101 in the local network can communicate with the cloud platform 111. When the computing resources in the HPC cluster are insufficient or the user has additional computing resource requirements, the task dispatching node 101 can apply for computing resources from the cloud platform 111. However, in the current technology, the cloud platform 111 only has the function of providing a general computing instance, and the general computing instance cannot meet the resource requirements of the HPC cluster based on the RoCE network.

[0066] To this end, the job processing method proposed by the embodiments of the present application can make the cloud platform 111 configure the cloud computing instance 112 with a RoCE function for the aforementioned HPC cluster, and enable the cloud computing instance 112 with the RoCE function to join the aforementioned HPC cluster as a local task processing node. Since the cloud computing instance 112 and the local task processing node 103 constitute a new HPC cluster, the task dispatching node 101 can assign tasks to the cloud computing instance 112 as a local task processing node. Therefore, the task dispatching node 101 does not need to transmit tasks and the corresponding task data to the cloud platform 111. Therefore, it is beneficial to improve the efficiency of high-performance computing.

[0067] For ease of understanding, the main flow of the job processing method proposed by the embodiments of the present application will be introduced first, as shown in the following table. Figure 2 As shown, the cloud platform and the cloud computing instance will perform the following steps:

[0068] 201. The cloud platform receives a resource application.

[0069] In the present embodiment, when the computing resources of the local network are insufficient or the user has additional computing resource requirements, the local network can apply for computing resources from the cloud platform. Therefore, the cloud platform can receive a resource application from the local network. That is, the resource application can be triggered by insufficient computing resources of the local network, or triggered based on the resource requirements of the user. The resource application is used to instruct the cloud platform to create a cloud computing instance with a RoCE function, and the RoCE function refers to the function of accessing a storage device over the operating system. The local network can be the aforementioned HPC cluster. Figure 1

[0070] ​Optionally, the resource request may also include other configuration information about the cloud computing instance, enabling the cloud platform to create a cloud computing instance that not only has RoCE functionality but is also compatible with the local network. For details, please refer to step 302 below.

[0071] 202. The cloud platform applies for and creates a cloud computing instance with RoCE functionality based on this resource.

[0072] Here, "cloud computing instance" refers to the computing resources configured by the cloud platform, which can also be understood as computing resources created by the cloud platform to support task processing on the local network. Different cloud computing instances can provide different computing capabilities, storage space, and network performance. Specifically, the cloud computing instance can be a bare metal server (BMS), that is, a physical server physically isolated from other users' servers, which combines the elasticity of virtual machines and the performance of physical machines. In addition, the cloud computing instance can also be a virtual machine or container created by a physical host, and the specifics are not limited here. In this embodiment and subsequent embodiments, only cloud computing instances are used as an example for description.

[0073] Once the cloud platform receives the aforementioned resource request, it will create a cloud computing instance with RoCE functionality based on the request. Specifically, the cloud platform can either directly create a cloud computing instance with RoCE functionality or create a regular cloud computing instance and then configure RoCE functionality for that regular instance; the specific choice is not limited here.

[0074] 203. Configure the cloud platform to connect the aforementioned cloud computing instance with RoCE functionality to the local network.

[0075] After the aforementioned cloud platform creates a cloud computing instance with RoCE functionality, the cloud platform also needs to configure the aforementioned cloud computing instance with RoCE functionality to access the local network so that the local network can communicate with the aforementioned cloud computing instance.

[0076] Optionally, when the local network is as described above Figure 1 When describing an HPC cluster, the local network can include task dispatch nodes, task data storage nodes, and compute nodes. When the aforementioned cloud computing instance with RoCE functionality accesses the local network, it can be assumed that the cloud computing instance can communicate with the aforementioned task dispatch nodes, task data storage nodes, and compute nodes.

[0077] It should be understood that the cloud platform can create only one cloud computing instance with RoCE function, or can create multiple cloud computing instances with RoCE function. When the cloud platform creates multiple cloud computing instances with RoCE function, each of the aforementioned multiple cloud computing instances will perform steps 204 and 205 described below, respectively.

[0078] 204. The cloud computing instance receives a task from the local network.

[0079] After the cloud computing instance accesses the local network, the cloud computing instance can receive a task from the local network, which is a high-performance computing task. The task can be sent to the cloud computing instance in the form of a message, an instruction or other forms. The task carries first indication information for indicating the data corresponding to the task. The cloud computing instance can access the data corresponding to the task located in the local network based on the first indication information carried in the task. Specifically, please refer to the detailed description in step 306 below.

[0080] 205. The cloud computing instance obtains the task data corresponding to the task from the task data storage node through RDMA according to the task, and performs task processing.

[0081] In this embodiment, since the cloud computing instance is configured with RoCE function, the cloud computing instance can obtain the task data corresponding to the task from the task data storage node through RDMA according to the first indication information carried in the task. Then, the cloud computing instance will perform task processing on the aforementioned task data. The task processing includes high-performance computing tasks, such as tasks in high-performance computing scenarios such as supercomputing centers and gene sequencing, or other tasks with large data volume, high requirements for computing performance, stability and real-time performance, etc.

[0082] In this embodiment, since the cloud platform can create a cloud computing instance with RoCE function according to the resource application, and access the aforementioned cloud computing instance to the local network, the cloud computing instance can receive a task from the local network through the operating system, and access the data corresponding to the task in the local network through the operating system, so as to process the data corresponding to the task in the aforementioned local network. Therefore, the cloud platform does not need to establish an HPC cluster on the cloud, and the local network does not need to transmit the task and the data corresponding to the task to the cloud platform. Therefore, it is beneficial to improve the efficiency of high-performance computing.

[0083] The job processing method will be further introduced based on the aforementioned embodiments in combination with the application scenarios and system architecture shown in Figure 1 As shown in Figure 3A Each node in the local network, the cloud platform and the cloud computing instance will perform the following steps:

[0084] 301. The task issuing node confirms that the number of tasks to be processed exceeds a threshold value.

[0085] In this embodiment, step 301 is an optional step.

[0086] In this embodiment, the task issuing node has a function of monitoring the number of tasks. Specifically, the task issuing node can count the number of tasks to be processed, which can be the number of tasks that the task issuing node has not assigned to the local task processing node, or the total amount of tasks that the task issuing node needs to process within a certain time range, which is not limited here. When the task issuing node detects that the number of tasks to be processed in the HPC cluster exceeds the threshold value, the task issuing node will send a resource application to the cloud platform. Thus, the cloud platform will perform step 302.

[0087] In some optional embodiments, the task issuing node includes an HPC controller and a bursting controller. The bursting controller monitors the number of tasks to be processed according to the job queue information controlled by the HPC controller, and triggers the step of sending a resource application to the cloud platform when the number of tasks to be processed reaches the threshold value. In the HPC scenario, the time range in which the number of tasks to be processed exceeds the threshold value is also referred to as the peak demand period. The process in which the task issuing node applies for resources to the cloud platform during the peak demand period is also referred to as cloud bursting. In this process, the aforementioned threshold value can be set by the task issuing node according to the computing power of the HPC cluster, which is not limited here.

[0088] 302. The cloud platform receives the resource application.

[0089] The resource application is used to instruct the cloud platform to create a cloud computing instance with RoCE function. The RoCE function refers to the function of accessing a storage device across an operating system. The cloud computing instance can be a bare-metal server, or a virtual machine or container created by a physical host, which is not limited here. Optionally, the resource application includes first identification information, which is used to indicate that the applied cloud computing instance needs to have RoCE function.

[0090] Specifically, the cloud platform receiving the aforementioned resource application can have multiple different implementations:

[0091] In an alternative embodiment, the resource application is triggered by a task issuing node when the task issuing node monitors that the number of tasks to be processed exceeds a threshold. That is, after the task issuing node performs step 301, the cloud platform can receive the resource application from the task issuing node.

[0092] In another alternative embodiment, the resource application can be triggered by a user-defined requirement. Specifically, a user can purchase or rent a service of configuring a cloud computing instance from the cloud platform. The cloud platform can provide an interface for configuring a cloud computing instance to the user through a client or a web browser, and when the user submits a configuration of a required cloud computing instance, the cloud platform can receive the resource application from the client or the web browser.

[0093] In actual applications, any of the above implementation manners can be adopted, and the specific embodiments are not limited.

[0094] In addition, the resource application further includes configuration information, which is used to indicate a basic configuration of a cloud computing instance required by the HPC cluster, and the basic configuration includes a host type required for composing the cloud computing instance, a number and capacity of hard disks, a type of network card, and an application type, etc.

[0095] In an alternative embodiment, the cloud platform includes a plurality of initial cloud computing instance templates, each initial cloud computing instance template has a fixed basic configuration, and each initial cloud computing instance template has a template number uniquely identifying the template. At this time, the configuration information in the resource application is the template number. At this time, after the cloud platform obtains the template number from the resource application, the cloud platform can know what kind of cloud computing instance needs to be configured to the HPC cluster.

[0096] In another alternative embodiment, the cloud platform does not have initial cloud computing instance templates, or the plurality of initial cloud computing instance templates in the cloud platform are inconsistent with the cloud computing instance required by the HPC cluster. At this time, the configuration information in the resource application contains detailed basic configurations. For example, the configuration information is that the type of the host is a bare metal server, 2 hard disks of 16 TB, 128G memory, a 10G ordinary network card, a data analysis application, and a data prediction application. At this time, the cloud platform can configure the initial cloud computing instance based on the configuration information.

[0097] Optionally, when the resource application indicates that a plurality of cloud computing instances need to be applied for, the resource application will contain the configuration information of each cloud computing instance in the plurality of cloud computing instances.

[0098] 303、The cloud platform creates a cloud computing instance with RoCE function according to the resource application.

[0099] When the cloud platform receives the aforementioned resource application, the cloud platform determines an initial cloud computing instance based on the aforementioned configuration information. Specifically, when the aforementioned configuration information adopts different embodiments, the cloud platform determines the initial cloud computing instance in different ways.

[0100] In an alternative embodiment, when the configuration information in the aforementioned resource application is a template number, the cloud platform selects an initial cloud computing instance corresponding to the template number from a plurality of initial cloud computing instances according to the template number.

[0101] In another alternative embodiment, when the configuration information in the aforementioned resource application contains detailed basic configurations, the cloud platform creates an initial cloud computing instance according to the aforementioned basic configurations.

[0102] When the cloud platform creates the aforementioned initial cloud computing instance, the cloud platform sets RoCE function for the aforementioned initial cloud computing instance by using a RoCE software package (software RoCE) to obtain a cloud computing instance with RoCE function.

[0103] In an alternative embodiment, the RoCE software package is pre-stored in a storage device of the cloud platform or a database managed by the cloud platform. The cloud platform obtains the RoCE software package from the aforementioned storage device or database and sends the RoCE software package to the aforementioned initial cloud computing instance. Then, the cloud platform triggers the RoCE software package to be installed in the initial cloud computing instance to obtain the cloud computing instance with RoCE function.

[0104] In another alternative embodiment, the RoCE software package can be written into the storage device of the initial cloud computing instance when the cloud platform configures the initial cloud computing instance. When it is necessary to configure the RoCE function for the initial cloud computing instance, the cloud platform triggers the initial cloud computing instance to install the RoCE software package.

[0105] Based on the aforementioned two alternative embodiments, in the process of triggering the initial cloud computing instance to install the RoCE software package, the cloud platform first triggers the boot disk in the cloud computing instance, and then copies the RoCE software package to the boot disk in the cloud computing instance. The boot disk is located in the startup program list in the computing node and is used to trigger the RoCE software package to start installation when the operating system in the cloud computing instance starts.

[0106] In this embodiment, a specific implementation of configuring the initial cloud computing instance without RoCE network card with RoCE function is proposed. The initial cloud computing instance is caused to simulate RoCE network card by installing RoCE software package, so that the initial cloud computing instance can use ordinary network card without RoCE network card. Therefore, the configuration cost of cloud computing instance can be saved, and the realizability of the scheme is improved.

[0107] In addition, the cloud platform also configures intermediate adaptation software in the cloud computing instance with RoCE function, which is used to modify the calling mode of the application in the cloud computing instance to remote direct memory access (RDMA) calling.

[0108] 304, the cloud platform sets the cloud computing instance with RoCE function to access the local network.

[0109] In this embodiment, after the cloud platform configures the cloud computing instance with RoCE function, in order to enable the cloud computing instance to serve the local HPC cluster, the cloud platform needs to set the cloud computing instance with RoCE function to access the local network (i.e. the local HPC cluster). Wherein, the local network is provided with task issuing node and task data storage node.

[0110] Specifically, in addition to the configuration information, the aforementioned resource application also includes HPC cluster information, which is used to indicate the address of each node in the HPC cluster and the connection between the HPC cluster and the gateway, so that the cloud platform accesses the cloud computing instance with RoCE function into the local HPC cluster according to the HPC cluster information.

[0111] Optionally, the HPC cluster information includes: internet protocol (IP) address of the task issuing node, port number of the task issuing node, user identification information (also known as tenant ID), universally unique identifier (UUID) of the bridge device, and port number of the bridge device. Wherein, the aforementioned bridge device is a level 2 bridge (L2BR) supporting RoCE protocol.

[0112] Optionally, when the cloud computing instance is a virtual machine, the HPC cluster information also includes virtual local area network (VLAN) range.

[0113] Specifically, the cloud platform accesses the cloud computing instances into the local HPC cluster according to the information of the HPC cluster. Wherein, the master node and the plurality of local task processing nodes are connected through a first switch, and the plurality of cloud computing instances configured by the cloud platform are connected through a second switch, and the first switch and the second switch are connected through a gateway and a bridging device. The bridging device can be the L2BR. Figure 3B For example, the local HPC cluster includes a local task processing node 1 (node1), a local task processing node 2 (node2) and a local task processing node 3 (node3), wherein each local task processing node is configured with a RoCE network interface controller (NIC) (i.e. RoCE network card). The plurality of local task processing nodes, the master node and the task data storage node in the local HPC cluster are connected through a first switch. In addition, the cloud platform creates a cloud computing instance 1 (node1'), a cloud computing instance 2 (node2') and a cloud computing instance 3 (node3'). Wherein, each cloud computing instance is configured with a normal network card, but each cloud computing instance runs a RoCE software, which can simulate the function of the RoCE network card. The plurality of cloud computing instances are connected through a second switch. The first switch and the second switch are connected through a gateway and a L2BR. Thus, the plurality of cloud computing instances are added to the local network (i.e. local HPC cluster) as task processing nodes.

[0114] It should be understood that the Figure 3B is only a logical connection diagram, and part of the physical gateway is not shown.

[0115] 305. The cloud platform sends a first notification to the master node.

[0116] In this embodiment, step 305 is an optional step.

[0117] Wherein, the first notification is used to indicate that the cloud computing instance is configured and has been accessed into the local network (i.e. local HPC cluster). After receiving the first notification, the master node will execute step 306.

[0118] In addition, when the cloud platform does not execute step 305, the master node can detect the idle computing resources in the HPC cluster. Since the cloud computing instance just accessed into the local network is in an idle state, when the master node detects that there is an idle computing resource, the master node will execute step 306.

[0119] 306、The task dispatching node sends the task to the cloud computing instance.

[0120] In this embodiment, after the task dispatching node receives the first notification, the task dispatching node can send the task in the task queue that is not assigned to the local task processing node to the cloud computing instance.

[0121] The task carries the first indication information, and the first indication information is used to indicate the data corresponding to the task. The cloud computing instance can access the data corresponding to the task in the local network based on the first indication information carried in the task.

[0122] Optionally, the first indication information includes a task identifier and / or an address of the data corresponding to the task.

[0123] The task identifier is used to uniquely identify one task in the task queue. For example, the task identifier can be a queue serial number or other characters or character strings. In addition, the data corresponding to the task also has the same task identifier. For example, the data corresponding to the task is stored in a task data storage node, and the head of the memory block where the data corresponding to the task is stored contains the aforementioned task identifier. In this implementation, when the cloud computing instance obtains the task identifier from the received task, the cloud computing instance can traverse the memory in the task data storage node. When it is detected that the head of a certain memory block contains the aforementioned task identifier, the cloud computing instance can obtain the data in the memory block, and thus the cloud computing instance can obtain the data corresponding to the task.

[0124] In addition, the first indication information can also include the address of the data corresponding to the task. The address of the data corresponding to the task can be a physical address or a logical address; can be the address of each memory block in the task data storage node, or can be the address of the memory block in a certain local task processing node in the local HPC cluster, and the specific implementation is not limited here. In this implementation, the cloud computing instance can directly obtain the data corresponding to the task from the address of the data corresponding to the task based on the first indication information.

[0125] Optionally, when the first indication information includes the address of the data corresponding to the task, the first indication information can further include a processing result address. The processing result address is used to store the processing result obtained by the cloud computing instance after processing the data corresponding to the task. The address of the data corresponding to the task and the processing result address can be addresses in the same node or device. For example, the address of the data corresponding to the task indicates a memory block in one of the task data storage nodes, and the processing result address indicates another memory block in the task data storage node. Of course, the address of the data corresponding to the task and the processing result address can also be addresses in different nodes or devices. For example, the address of the data corresponding to the task indicates a memory block in one of the task data storage nodes, and the processing result address indicates a memory block in the local task processing node. The specific implementation is not limited here.

[0126] 307. The cloud computing instance obtains the task data corresponding to the task from the task data storage node through remote direct memory access (RDMA) according to the task, and performs task processing.

[0127] In this embodiment, after receiving the task, the cloud computing instance obtains the task data corresponding to the task from the task data storage node through RDMA according to the first indication information carried by the task. Optionally, if the task needs to use data generated by a certain computing node, the cloud computing instance can also access the data in the storage device in the computing node without operating system. Specifically, when the first indication information is different, the cloud computing instance obtains the task data through RDMA in different ways. For details, please refer to the related description in step 306, which will not be repeated here.

[0128] Then, the cloud computing instance performs task processing on the task data. The task processing includes high-performance computing tasks, such as tasks in high-performance computing scenarios such as supercomputing centers and gene sequencing, or other tasks with large data volume, high requirements for computing performance, stability, and real-time performance.

[0129] Optionally, the task data storage node can be independent of the computing nodes in the local HPC cluster, for example, the task data storage node is a database in the local HPC cluster. In addition, the task data storage node can also be integrated with the task issuing node. For example, the task data storage node can be located in the task issuing node as a storage device in the task issuing node. At this time, the task data storage node can not only store the data corresponding to the task, but also store the task queue formulated by the task issuing node.

[0130] In the embodiment, when the cloud computing instance with RoCE function configured by the cloud platform enters the local network, the cloud computing instance joins the local HPC cluster as a computing node. Therefore, the cloud computing instance can receive tasks from the task issuing node, and can access the data corresponding to the tasks in the task data storage node through the operating system, and then the cloud computing instance can process the data corresponding to the tasks. Therefore, the task issuing node does not need to transmit the tasks and the data corresponding to the tasks to the cloud platform. Therefore, it is beneficial to improve the efficiency of high-performance computing.

[0131] Based on the foregoing embodiments, as shown in Figure 3A When the local HPC cluster does not need to use the foregoing cloud computing instance, the local HPC cluster will revoke the cloud computing instance by the following steps.

[0132] 308. The task issuing node sends a resource revocation request to the cloud platform.

[0133] The resource revocation request can request to revoke all cloud computing instances applied by the cloud platform, or can only revoke a certain cloud computing instance.

[0134] When the resource revocation request is used to request to revoke all cloud computing instances applied by the cloud platform, the resource revocation request can only carry the identification information of the user. When the resource revocation request is used to request to revoke a certain cloud computing instance, the resource revocation request contains second identification information, which is used to indicate the cloud computing instance to be revoked. The second identification information can be a template number corresponding to the cloud computing instance, or a string uniquely identifying the foregoing cloud computing instance. Optionally, the second identification information can be set by the cloud platform when creating the cloud computing instance, or can be carried by the task issuing node in the foregoing resource application, which is not limited here.

[0135] It should be understood that the task issuing node can perform the step 308 when confirming that the tasks of the HPC cluster have been completed. In addition, if the user does not need to use the cloud computing instance, for example, the user's rental period expires, or the user suspends the rental of the cloud computing instance. At this time, the task issuing node can also trigger the above step 308.

[0136] 309. The cloud platform revokes the cloud computing instance according to the resource revocation request.

[0137] When the cloud platform receives the foregoing resource revocation request, the cloud platform will revoke one or more cloud computing instances corresponding to the second identification information according to the second identification information in the resource revocation request.

[0138] Optionally, when the cloud platform revokes the cloud computing instance, the cloud platform also sends a second notification to the task issuing node, and the second notification is used to inform that the cloud computing instance corresponding to the second identification information has been revoked.

[0139] In the embodiment, when the task is completed or the task does not need to be executed, the task issuing node can send a resource revoking request to the cloud platform, so that the cloud platform revokes the cloud computing instance according to the revoking request. In this scheme, the task issuing node can revoke the cloud computing instance, and the cloud platform can also allocate the cloud computing instance to other clusters. Therefore, the cloud computing instance configured by the cloud platform can be more flexible, and the utilization rate of the cloud computing instance can be improved.

[0140] The device involved in the job processing method in the embodiment of the application will be introduced below:

[0141] As shown in Figure 4 , a structural schematic diagram of a cloud platform 40 is provided in the embodiment of the application. The cloud platform 40 can be a server, a large computing device or a large management device, and the specific implementation is not limited here. The cloud platform 40 is used to allocate a cloud computing instance to a task issuing node. Figure 2 and Figure 3A The cloud platform in the corresponding method embodiment can be based on the structure of the cloud platform 40 as shown in Figure 4 .

[0142] The cloud platform 40 includes at least one processor 401 and at least one memory 402. It should be understood that Figure 4 only one processor 401 and one memory 402 are shown.

[0143] The processor 401 can be a general central processing unit (CPU), a microprocessor, a network processor (NP), or a specific application integrated circuit (ASIC), or one or more integrated circuits for controlling the execution of programs of the present application. The aforementioned processor 401 can be a single-CPU processor or a multi-CPU processor. The processor 401 can refer to one or more devices, circuits, and / or processing cores for processing data (e.g., computer program instructions). In addition, the processor 401 can be a separate semiconductor chip or integrated with other circuits as a semiconductor chip, for example, it can be integrated with other circuits (such as coding and decoding circuits, hardware acceleration circuits, or various bus and interface circuits) to form a system-on-a-chip (SoC), or it can be integrated as a built-in processor of an application specific integrated circuit (ASIC), which can be packaged separately or packaged together with other circuits.

[0144] In addition, the aforementioned memory 402 can be a read-only memory (ROM), other types of static storage devices that can store static information and instructions, a random access memory (RAM), other types of dynamic storage devices that can store information and instructions, an electrically erasable programmable read-only memory (EEPROM), or the like, which is not limited here. The memory 402 can exist independently, but is connected to the aforementioned processor 401. Optionally, the memory 402 can be integrated with the aforementioned processor 401. For example, it can be integrated in one or more chips.

[0145] Furthermore, the memory 402 is also used to store program code that executes the technical solutions of the embodiments of this application. The aforementioned program code can be controlled and executed by the processor 401, and the various types of computer program code being executed can also be considered as drivers for the processor 401. Therefore, the processor 401 can analyze received resource requests, set up a cloud computing instance with RoCE functionality according to the resource request, and configure the cloud computing instance to access the local network. Optionally, the processor 401 can also create an initial cloud computing instance and configure RoCE functionality for the initial cloud computing instance. Optionally, the processor 401 can also analyze resource revocation requests and revoke the cloud computing instance configured for the HPC cluster according to the resource revocation request.

[0146] Optionally, the cloud platform 40 also includes a communication interface 403 for communicating with other servers or network devices, enabling the cloud platform to receive instructions or data from other devices. For example, the communication interface 403 can receive resource requests or resource cancellation requests from task transceiver devices. The communication interface 403 is also used to send data or instructions to other devices. For example, the communication interface 403 can send a RoCE software package to an initial cloud computing instance, allowing the initial cloud computing instance to install the RoCE software program based on the RoCE software package.

[0147] like Figure 5 The diagram shown is a structural schematic of a cloud platform 50 provided in an embodiment of this application. The cloud platform 50 can be a server, a large computing device, or a large management device; no specific limitation is made here. (The preceding text is incomplete and requires further context.) Figure 2 and Figure 3A The cloud platforms in the corresponding method embodiments can all be based on Figure 5 The structure of the cloud platform 50 shown is illustrated.

[0148] The cloud platform 50 includes multiple functional modules. These modules can be integrated into a single processing unit, exist as separate physical units, or be integrated into a single unit. The integrated unit can be implemented in hardware or as a software functional unit.

[0149] Specifically, the cloud platform includes 50 components:

[0150] The receiving module 501 is used to receive a resource request, which is used to instruct the cloud platform to create a cloud computing instance with Remote Direct Memory Access (RoCE) functionality based on Converged Ethernet.

[0151] The resource configuration module 502 is configured to set the cloud computing instance with the RoCE function according to the resource application, and set the cloud computing instance to access the local network. The local network is provided with a task issuing node and a task data storage node. The task issuing node is configured to send a task to the cloud computing instance, and the task data storage node is configured to provide the cloud computing instance with task data corresponding to the task through remote direct memory access (RDMA).

[0152] In the embodiment, the cloud platform 50 can create the cloud computing instance with the RoCE function according to the resource application, and access the cloud computing instance to the local network, so that the cloud computing instance can receive the task from the local network and access the data corresponding to the task in the local network through the operating system, to process the data corresponding to the task in the local network. Therefore, the cloud platform does not need to establish an HPC cluster on the cloud, and the local network does not need to transmit the task and the data corresponding to the task to the cloud platform. Therefore, the efficiency of high-performance computing is improved.

[0153] In another optional implementation, the receiving module 501 is specifically configured to receive the resource application from the task issuing node when the task issuing node confirms that the number of tasks to be processed exceeds a threshold.

[0154] In another optional implementation, the receiving module 501 is further configured to receive a resource revocation request from the task issuing node, the resource revocation request being used to indicate that the task is completed or the task does not need to be executed, and the resource configuration module 502 is further configured to revoke the cloud computing instance according to the resource revocation request.

[0155] In another optional implementation, the resource configuration module 502 is specifically configured to obtain a RoCE software package, control the transceiver to send the RoCE software package to an initial cloud computing instance, and trigger the RoCE software package to be installed in the initial cloud computing instance to obtain the cloud computing instance with the RoCE function.

[0156] In another optional implementation, the resource configuration module 502 is further configured to create an initial cloud computing instance according to the configuration information, or find an initial cloud computing instance corresponding to the configuration information from a plurality of initial cloud computing instances according to the configuration information.

[0157] In the above embodiments, each functional module in the cloud platform 50 can be implemented by software, hardware, firmware or any combination thereof, in whole or in part. When implemented by software, each functional module in the cloud platform 50 can be implemented in the form of a computer program product, in whole or in part. At this time, if each functional module in the cloud platform 50 is implemented in the form of a software function unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solutions of the present application, in essence or the part that contributes to the prior art, or the whole or part of the technical solutions can be embodied in the form of a software product. For example, as a computer software product stored in a storage medium, the computer software product includes a number of instructions to make a computer device (which can be a personal computer, a server, or a network device, etc.) execute all or part of the steps of the method described in each embodiment of the present application. The aforementioned storage medium includes a U disk, a mobile hard disk, a read-only memory (ROM), a random access memory (RAM), a magnetic disk or an optical disk, and various program code storage media.

[0158] Those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working processes of the above-described system, device and module can refer to the corresponding processes in the foregoing method embodiments, which will not be described here.

[0159] The above embodiments are only used to illustrate the technical solutions of the present application, rather than limit them; although the present application has been described in detail with reference to the foregoing embodiments, those of ordinary skill in the art should understand that they can still modify the technical solutions recorded in the foregoing embodiments, or make equivalent replacements for some technical features; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions of the embodiments of the present application.

Claims

1. A job processing method characterized by comprising: The cloud platform receives a resource application under a preset condition, the resource application being used to instruct the cloud platform to create a cloud computing instance with a remote direct memory access over converged Ethernet (RoCE) function, and the preset condition including that computing resources of a local network are insufficient to support high-performance computing. The cloud platform creates the cloud computing instance with the RoCE function according to the resource application, and sets the cloud computing instance to access the local network. The local network is provided with a task issuing node and a task data storage node, and the method further includes: The cloud computing instance receives a task sent by the task issuing node. The cloud computing instance acquires task data corresponding to the task from the task data storage node through remote direct memory access (RDMA) according to the task, and performs task processing. Before the cloud platform receives the resource application under the preset condition, the method further includes:

2. The method of claim 1, wherein, The task issuing node confirms that the number of tasks to be processed exceeds a threshold value. The method further includes:

3. The method according to any one of claims 1 to 2, characterized in that, The cloud platform receives a resource revocation request sent by the task issuing node, the resource revocation request being used to indicate that the task has been completed or the task does not need to be executed. The cloud platform revokes the cloud computing instance according to the resource revocation request. The cloud platform creates the cloud computing instance with the RoCE function according to the resource application, including:

4. The method according to any one of claims 1 to 2, characterized in that, The cloud platform acquires an RoCE software package. The cloud platform sends the RoCE software package to an initial cloud computing instance, and triggers the RoCE software package to be installed in the initial cloud computing instance to obtain the cloud computing instance with the RoCE function. The resource application includes configuration information of the cloud computing instance.

5. The method of claim 4, wherein, The method further includes: The cloud platform creates an initial cloud computing instance according to the configuration information. Alternatively, The cloud platform selects a cloud computing instance corresponding to the configuration information from a plurality of cloud computing instances according to the configuration information. The cloud computing instance is a virtual machine, a container or a bare metal server.

6. The method according to any one of claims 1 to 2, characterized in that, The cloud platform receives a resource application under a preset condition, the resource application being used to instruct the cloud platform to create a cloud computing instance with a remote direct memory access over converged Ethernet (RoCE) function, and the preset condition including that computing resources of a local network are insufficient to support high-performance computing.

7. A cloud platform, characterized by The cloud platform creates the cloud computing instance with the RoCE function according to the resource application, and sets the cloud computing instance to access the local network. The local network is provided with a task issuing node and a task data storage node, and the method further includes: The cloud computing instance receives a task sent by the task issuing node. The cloud computing instance acquires task data corresponding to the task from the task data storage node through remote direct memory access (RDMA) according to the task, and performs task processing.

8. The cloud platform of claim 7, wherein The receiving module is further configured to receive a resource revocation request sent by the task issuing node, the resource revocation request being used to indicate that the task has been completed or the task does not need to be executed. The resource configuration module is further configured to revoke the cloud computing instance according to the resource revocation request.

9. The cloud platform of claim 7 or 8, wherein, The resource configuration module is specifically configured to: obtain a RoCE software package; send the RoCE software package to an initial cloud computing instance, and trigger the RoCE software package to be installed in the initial cloud computing instance to obtain the cloud computing instance with the RoCE function.

10. The cloud platform of claim 9, wherein, The resource application includes configuration information of the cloud computing instance. The resource configuration module is further configured to: create an initial cloud computing instance according to the configuration information; or select a cloud computing instance corresponding to the configuration information from a plurality of cloud computing instances according to the configuration information.

11. A cloud platform, characterized by A processor and a memory are included, the memory stores program instructions, and the processor executes the program instructions to implement the method in any one of claims 1, 3-6.

12. A job processing system characterized by comprising: It includes: a cloud platform configured to receive a resource application under a preset condition, create a cloud computing instance with a RoCE function according to the resource application, and set the cloud computing instance to access a local network, the resource application being configured to instruct the cloud platform to create a cloud computing instance with a RoCE function based on a converged Ethernet, and the preset condition including that a computing resource of the local network is insufficient to support high-performance computing; the cloud computing instance is configured to receive a task from the local network, obtain task data corresponding to the task in the local network through remote direct memory access (RDMA), and perform task processing; a task distribution node and a task data storage node are arranged in the local network; the cloud computing instance is specifically configured to receive a task sent by the task distribution node, obtain task data corresponding to the task from the task data storage node through remote direct memory access (RDMA) according to the task, and perform task processing.

13. The work processing system according to claim 12, characterized by The task distribution node is further configured to confirm that the number of tasks to be processed exceeds a threshold.

14. The job processing system of any one of claims 12-13, wherein: the cloud platform is further configured to receive a resource revocation request sent by the task distribution node, the resource revocation request being configured to indicate that the task has been completed or the task does not need to be executed; the cloud platform is further configured to revoke the cloud computing instance according to the resource revocation request.

15. The job processing system according to any one of claims 12 to 13, characterized by, The cloud platform is specifically configured to: obtain a RoCE software package; send the RoCE software package to an initial cloud computing instance, and trigger the RoCE software package to be installed in the initial cloud computing instance to obtain the cloud computing instance with the RoCE function.

16. The work processing system according to claim 15, characterized by The resource application includes configuration information of the cloud computing instance. The cloud platform is further configured to: create an initial cloud computing instance according to the configuration information; or select a cloud computing instance corresponding to the configuration information from a plurality of cloud computing instances according to the configuration information.

17. The job processing system according to any one of claims 12 to 13, characterized by, The cloud computing instance is a virtual machine, a container, or a bare metal server.

Images