WO2020108536A1 - 一种虚拟网络资源分配方法、系统及电子设备 - Google Patents

一种虚拟网络资源分配方法、系统及电子设备 Download PDF

Info

Publication number
WO2020108536A1
WO2020108536A1 PCT/CN2019/121297 CN2019121297W WO2020108536A1 WO 2020108536 A1 WO2020108536 A1 WO 2020108536A1 CN 2019121297 W CN2019121297 W CN 2019121297W WO 2020108536 A1 WO2020108536 A1 WO 2020108536A1
Authority
WO
WIPO (PCT)
Prior art keywords
virtual
graph
network information
network
critical path
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Ceased
Application number
PCT/CN2019/121297
Other languages
English (en)
French (fr)
Inventor
叶志勇
王洋
须成忠
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Shenzhen Institute of Advanced Technology of CAS
Original Assignee
Shenzhen Institute of Advanced Technology of CAS
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Shenzhen Institute of Advanced Technology of CAS filed Critical Shenzhen Institute of Advanced Technology of CAS
Publication of WO2020108536A1 publication Critical patent/WO2020108536A1/zh
Anticipated expiration legal-status Critical
Ceased legal-status Critical Current

Links

Images

Classifications

    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
    • H04L41/00—Arrangements for maintenance, administration or management of data switching networks, e.g. of packet switching networks
    • H04L41/08—Configuration management of networks or network elements
    • H04L41/0896—Bandwidth or capacity management, i.e. automatically increasing or decreasing capacities
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
    • H04L41/00—Arrangements for maintenance, administration or management of data switching networks, e.g. of packet switching networks
    • H04L41/08—Configuration management of networks or network elements
    • H04L41/0893—Assignment of logical groups to network elements

Definitions

  • the present application belongs to the field of computer virtualization technology, and particularly relates to a virtual network resource allocation method, system and electronic equipment.
  • virtualization technology is often used to improve the resource utilization rate of the data center.
  • the so-called virtualization technology refers to the technology of being able to run multiple virtual machines on one physical machine, and the virtual machine is almost indistinguishable from the physical machine in terms of users, application software and even the operating system. Although this can greatly improve resource utilization, it also leads to the problem of resource allocation between different virtual machines. Therefore, how to effectively allocate virtual resources and maximize resource utilization has been widely concerned and studied.
  • the present application provides a virtual network resource allocation method, system, and electronic device, which are intended to solve at least to a certain extent one of the above technical problems in the prior art.
  • a virtual network resource allocation method includes the following steps:
  • Step a Generate a task graph based on the network information of all virtual machines on the physical machine, and use a multi-level graph partition algorithm to divide the task graph to obtain k disjoint subsets;
  • Step b Calculate the key path with the longest completion time in each split subset to obtain the virtual machine at the key path node;
  • Step c Allocate a single root input and output virtualized virtual device to the virtual machine at the critical path node.
  • the technical solution adopted in the embodiment of the present application further includes: before step a, the method further includes: collecting network information of all virtual machines on each physical machine; the way of collecting network information is: each virtual machine reads the UNIX operating system as The user accesses the pseudo file system provided by the kernel to obtain network information.
  • the network information obtained by the virtual machine is placed in a storage system.
  • Each physical machine storage system reads the network information of all virtual machines and uploads the network information to the host.
  • the technical solution adopted in the embodiment of the present application further includes: in the step a, the task graph is generated according to the network information of all virtual machines on the physical machine, and the task graph is divided using a multi-level graph partitioning algorithm specifically: The host receives the network information uploaded by all physical machines, generates a task map with parameters according to the network information, and divides the task map using a multi-level recursive binary division method.
  • the multi-level recursive binary division method is used to divide the task graph specifically: the multi-level recursive binary division method includes three stages of coarsening, initial division, and refinement.
  • the coarsening stage generates a series of aggregation graphs with lower complexity after inputting the task graph, and reduces the complexity of the task graph by the vertices of the largest adjacent pair of the aggregation graph, when the complexity of the task graph drops to a preset
  • the coarsening phase ends when the number of vertices is reached; the initial division phase uses a heuristic algorithm to initially divide the task graph, and the thinning phase gradually restores the task graph after the initial division.
  • the technical solution adopted in the embodiment of the present application further includes:
  • the calculation formula of the critical path is:
  • path refers to a certain path in the task graph
  • PathSet refers to the entire set of paths in the task graph
  • PathCompTime refers to the execution time of all nodes in the path
  • symbol P refers to any path
  • CP refers to Critical Path.
  • a virtual network resource allocation system includes a physical machine and a host, and the host includes:
  • Multi-level graph partitioning module used to generate a task graph based on the network information of all virtual machines on the physical machine, and using a multi-level graph partitioning algorithm to divide the task graph to obtain k disjoint subsets that are mutually disjoint;
  • Critical path calculation module used to calculate the critical path with the longest completion time in each split subset to obtain the virtual machine located in the critical path node;
  • Network resource allocation module used to allocate a single root input and output virtualized virtual device to the virtual machine at the critical path node.
  • the technical solution adopted in the embodiments of the present application further includes: the physical machine includes a network resource monitoring module, and the network resource monitoring module is used to collect network information of all virtual machines on the physical machine and upload the collected network information to network resource management Module; the network information is collected by: each virtual machine obtains network information by reading the pseudo file system provided by the UNIX operating system for the user to access the kernel, and the network information obtained by the virtual machine is put into the storage system, each physical The machine storage system reads the network information of all virtual machines and uploads the network information to the network resource management module.
  • the technical solution adopted in the embodiment of the present application further includes: the host further includes a network resource management module, the network resource management module is configured to receive network information uploaded by all physical machines, and input the network information into the multi-level graph dividing module;
  • the multi-level graph division module generates a task graph with parameters according to the network information, and uses a multi-level recursive binary division method to divide the task graph.
  • the multi-level recursive binary division method is used to divide the task graph specifically: the multi-level recursive binary division method includes three stages of coarsening, initial division, and refinement.
  • the coarsening stage generates a series of aggregation graphs with a lower complexity after inputting the task graph, and reduces the complexity of the task graph by the vertices of the largest adjacent pair of the aggregation graph, when the complexity of the task graph drops to a preset
  • the coarsening phase ends when the number of vertices is reached; the initial division phase uses a heuristic algorithm to initially divide the task graph, and the thinning phase gradually restores the task graph after the initial division.
  • the technical solution adopted in the embodiment of the present application further includes: the calculation formula of the critical path is:
  • path refers to a certain path in the task graph
  • PathSet refers to the entire set of paths in the task graph
  • PathCompTime refers to the execution time of all nodes in the path
  • symbol P refers to any path
  • CP refers to Critical Path.
  • an electronic device including:
  • At least one processor At least one processor
  • a memory communicatively connected to the at least one processor; wherein,
  • the memory stores instructions executable by the one processor, and the instructions are executed by the at least one processor to enable the at least one processor to perform the following operations of the virtual network resource allocation method described above:
  • Step a Generate a task graph based on the network information of all virtual machines on the physical machine, and use a multi-level graph partition algorithm to divide the task graph to obtain k disjoint subsets;
  • Step b Calculate the critical path with the longest completion time in each split subset to obtain the virtual machine located in the critical path node;
  • Step c Allocate a single root input and output virtualized virtual device to the virtual machine at the critical path node.
  • the virtual network resource allocation method, system and electronic device of the embodiments of the present application are optimized by a combination of paravirtualization technology and single root input and output virtualization technology Network, by default, each virtual machine is assigned a para-virtualized virtual network card, and the single-input virtualized virtual device is only assigned to the virtual machines located in the critical path nodes after the multi-level graph is divided, so as to improve the overall network bandwidth The purpose of utilization and reducing the overall execution time of parallel applications.
  • FIG. 1 is a flowchart of a virtual network resource allocation method according to an embodiment of this application.
  • FIG. 2 is a schematic structural diagram of a virtual network resource allocation system according to an embodiment of the present application.
  • FIG. 3 is a schematic structural diagram of a hardware device of a virtual network resource allocation method according to an embodiment of the present application.
  • the virtual network resource allocation method of the embodiment of the present application uses a combination of para-virtualization technology and single-root input/output virtualization technology (SR-IOV, Single-Root I/O Virtualization) to optimize
  • SR-IOV single-root input/output virtualization technology
  • VF Virtual Function
  • Critical Path Critical Path
  • FIG. 1 is a flowchart of a virtual network resource allocation method according to an embodiment of the present application.
  • the virtual network resource allocation method in the embodiment of the present application includes the following steps:
  • Step 100 Collect network information of all virtual machines on each physical machine, and upload the collected network information to the host;
  • step 100 the network information is collected as follows: each virtual machine obtains network information by reading the pseudo file system /proc/net/dev provided by the UNIX operating system for the user to access the kernel, and the network information obtained by the virtual machine is stored In the system (XenStore), each physical machine is provided with a network resource monitoring module.
  • the network resource monitoring module reads the network information of each virtual machine through XenStore and sends it to the network resource management module.
  • Step 200 Receive the network information of all virtual machines on each physical machine through the host, generate a task graph with parameters, and use a multi-level graph division algorithm to divide the task graph to obtain k disjoint subsets that are not intersecting;
  • the host includes a network resource management module and a multi-level graph dividing module.
  • the network resource management module is responsible for receiving network information of all virtual machines on each physical machine, and inputting the network information into the multi-level graph dividing module to generate Task graph
  • multi-level graph partitioning module adopts multi-level recursive bipartite partition method to partition the task graph, and obtain k disjoint subsets.
  • the tasks corresponding to different partition subsets are allocated to different physical machines for calculation.
  • the task graph is divided into multi-level graphs, so as to balance the calculation amount between each physical machine and minimize the total communication amount between each physical machine.
  • the multi-level recursive binary division method includes three stages of coarsening, initial division and refinement.
  • coarsening stage after inputting the task graph, a series of aggregation graphs with smaller complexity will be generated.
  • the purpose is to reduce the complexity of the task graph by the vertices of the largest adjacent pair of the aggregation graph.
  • the roughening phase ends.
  • the initial division stage because the complexity of the task graph is already very small, a classic simple heuristic algorithm such as the K-L algorithm is used to initially divide the task graph.
  • refinement stage the task graph after the initial division is gradually restored, and the vertices that are aggregated are moved between different segmentation subsets to improve the division effect of the task graph.
  • Step 300 Calculate the critical path with the longest completion time in each split subset to obtain the virtual machine located on the critical path node on each physical machine;
  • step 300 since the virtual machine communication cost between the same physical machines can be regarded as zero, the critical path calculation after the task graph is divided only calculates the communication cost between different divided subsets. Specifically, the calculation formula of the critical path is:
  • path refers to a certain path in the task graph
  • PathSet refers to the entire set of paths in the task graph
  • PathCompTime refers to the execution time of all nodes in the path
  • the symbol P refers to any path
  • CP refers to the critical path.
  • Step 400 Assign a single root input and output virtualized virtual device to the virtual machines on the critical path nodes on each physical machine;
  • the network resource allocation method is to allow each virtual machine to be assigned a paravirtualized virtual network card by default, and a single root input/output virtualized virtual device is only allocated to the multi-level graph after the division.
  • the multi-network card binding technology is used to transparently switch between the para-virtualized virtual network card and the single-root virtualized virtual device.
  • the single-root virtualized virtual device is set as the main network card, and the para-virtualized virtual network card It is set as the slave network card.
  • the master network card is valid, the virtual machine uses the master network card to communicate. Only when the master network card is invalid, the slave network card is used to communicate, thereby reducing network overhead.
  • Step 500 Determine whether all the split subsets have been executed. If all the split subsets have been executed, go to step 600; otherwise, return to step 300 to continue to calculate the critical path calculation of the next split subset until all the split subsets have been executed;
  • Step 600 The virtual network resource allocation ends.
  • FIG. 2 is a structural diagram of a virtual network resource allocation system according to an embodiment of the present application.
  • the virtual network resource allocation system according to an embodiment of the present application includes a physical machine and a host.
  • the physical machine includes a network resource monitoring module.
  • the host includes a network resource management module, a multi-level graph dividing module, a critical path calculation module, a network resource allocation module, and task execution judgment Module.
  • Network resource monitoring module used to collect network information of all virtual machines on each physical machine and upload the collected network information to the network resource management module; where the network information is collected as follows: each virtual machine reads the UNIX operating system The user accesses the pseudo file system /proc/net/dev provided by the kernel to obtain network information. The network information obtained by the virtual machine is placed in XenStore. The network resource monitoring module reads the network information of each virtual machine through XenStore and sends it to the network. Resource management module.
  • Network resource management module used to receive network information of all virtual machines on each physical machine and input the network information into the multi-level graph division module;
  • Multi-level graph division module used to generate a task graph with parameters according to the input network information, and use a multi-level graph partition algorithm to divide the task graph to obtain k disjoint subsets; among them, multi-level graph
  • the graph division module uses a multi-level recursive binary division method to divide the task graph to obtain k disjoint subsets.
  • the tasks corresponding to different subsets are allocated to different physical machines for calculation.
  • the task graph is divided into multi-level graphs, so as to balance the calculation amount between each physical machine and minimize the total communication amount between each physical machine.
  • the multi-level recursive binary division method includes three stages of coarsening, initial division and refinement.
  • coarsening stage after inputting the task graph, a series of aggregation graphs with smaller complexity will be generated.
  • the purpose is to reduce the complexity of the task graph by the vertices of the largest adjacent pair of the aggregation graph.
  • the complexity of the task graph is reduced to only a few hundred At the top of the vertices.
  • a classic simple heuristic algorithm such as the K-L algorithm is used to initially divide the task graph.
  • refinement stage the task graph after the initial division is gradually restored, and the vertices that are aggregated are moved between different segmentation subsets to improve the division effect of the task graph.
  • Critical path calculation module used to calculate the critical path with the longest completion time in each divided subset to obtain the virtual machine located on the critical path node on each physical machine; where, due to the virtual machine communication overhead between the same physical machine can be regarded as zero Therefore, the critical path calculation after the task graph is divided only calculates the communication overhead between different divided subsets.
  • the calculation formula of the critical path is:
  • path refers to a certain path in the task graph
  • PathSet refers to the entire set of paths in the task graph
  • PathCompTime refers to the execution time of all nodes in the path
  • the symbol P refers to any path
  • CP refers to the critical path.
  • Network resource allocation module used to allocate single root input and output virtualized virtual devices to virtual machines on critical path nodes on each physical machine; wherein, the network resource allocation method of the embodiment of the present application is to allow each virtual machine to be allocated by default Para-virtualized virtual network cards, and single-input virtualized virtual devices are only assigned to virtual machines located at critical path nodes after multi-level graph partitioning.
  • the multi-network card binding technology is used to transparently switch between the para-virtualized virtual network card and the single-root virtualized virtual device.
  • the single-root virtualized virtual device is set as the main network card, and the para-virtualized virtual network card It is set as the slave network card.
  • the master network card is valid, the virtual machine uses the master network card to communicate. Only when the master network card is invalid, the slave network card is used to communicate, thereby reducing network overhead.
  • Task execution judgment module used to determine whether all the split subsets have been executed. If all the split subsets have been executed, the virtual network resource allocation ends; otherwise, continue to calculate the critical path calculation of the next split subset through the critical path calculation module. Until all split subsets are executed.
  • the device includes one or more processors and memory. Taking a processor as an example, the device may further include: an input system and an output system.
  • the processor, the memory, the input system, and the output system may be connected by a bus or other means.
  • a bus In FIG. 3, the connection by a bus is used as an example.
  • the memory can be used to store non-transitory software programs, non-transitory computer executable programs, and modules.
  • the processor runs non-transitory software programs, instructions, and modules stored in the memory to execute various functional applications and data processing of the electronic device, that is, to implement the processing methods of the foregoing method embodiments.
  • the memory may include a storage program area and a storage data area, where the storage program area may store an operating system and application programs required by at least one function; the storage data area may store data, and the like.
  • the memory may include a high-speed random access memory, and may also include a non-transitory memory, such as at least one magnetic disk storage device, a flash memory device, or other non-transitory solid-state storage devices.
  • the memory optionally includes memories remotely located with respect to the processor, and these remote memories may be connected to the processing system via a network. Examples of the above network include but are not limited to the Internet, intranet, local area network, mobile communication network, and combinations thereof.
  • the input system can receive input digital or character information, and generate signal input.
  • the output system may include display devices such as display screens.
  • the one or more modules are stored in the memory, and when executed by the one or more processors, perform the following operations of any of the foregoing method embodiments:
  • Step a Generate a task graph based on the network information of all virtual machines on the physical machine, and use a multi-level graph partition algorithm to divide the task graph to obtain k disjoint subsets;
  • Step b Calculate the critical path with the longest completion time in each split subset to obtain the virtual machine located in the critical path node;
  • Step c Allocate a single root input and output virtualized virtual device to the virtual machine at the critical path node.
  • the above-mentioned products can execute the method provided in the embodiments of the present application, and have function modules and beneficial effects corresponding to the execution method.
  • function modules and beneficial effects corresponding to the execution method For technical details that are not described in detail in this embodiment, refer to the method provided in the embodiments of the present application.
  • An embodiment of the present application provides a non-transitory (non-volatile) computer storage medium that stores computer-executable instructions, and the computer-executable instructions can perform the following operations:
  • Step a Generate a task graph based on the network information of all virtual machines on the physical machine, and use a multi-level graph partition algorithm to divide the task graph to obtain k disjoint subsets;
  • Step b Calculate the critical path with the longest completion time in each split subset to obtain the virtual machine located in the critical path node;
  • Step c Allocate a single root input and output virtualized virtual device to the virtual machine at the critical path node.
  • An embodiment of the present application provides a computer program product.
  • the computer program product includes a computer program stored on a non-transitory computer-readable storage medium.
  • the computer program includes program instructions. When the program instructions are executed by a computer To cause the computer to perform the following operations:
  • Step a Generate a task graph based on the network information of all virtual machines on the physical machine, and use a multi-level graph partition algorithm to divide the task graph to obtain k disjoint subsets;
  • Step b Calculate the critical path with the longest completion time in each split subset to obtain the virtual machine located in the critical path node;
  • Step c Allocate a single root input and output virtualized virtual device to the virtual machine at the critical path node.
  • the virtual network resource allocation method, system, and electronic device of the embodiments of the present application use a combination of paravirtualization technology and single-root input-output virtualization technology to optimize the network.
  • each virtual machine is assigned a paravirtualized virtual network card
  • single-input virtualized virtual devices are only allocated to virtual machines located at critical path nodes after multi-level graph partitioning, in order to improve the overall network bandwidth utilization and reduce the overall execution time of parallel applications.

Landscapes

  • Engineering & Computer Science (AREA)
  • Computer Networks & Wireless Communication (AREA)
  • Signal Processing (AREA)
  • Data Exchanges In Wide-Area Networks (AREA)
  • Information Transfer Between Computers (AREA)

Abstract

本申请涉及一种虚拟网络资源分配方法、系统及电子设备。所述方法包括：步骤a：根据物理机上所有虚拟机的网络信息生成任务图，并采用多级图划分算法对所述任务图进行划分，得到k个互不相交的分割子集；步骤b：计算各个分割子集中具有最长完成时间的关键路径，得到位于关键路径节点的虚拟机；步骤c：为所述处于关键路径节点的虚拟机分配单根输入输出虚拟化的虚拟设备。本申请采用半虚拟化技术与单根输入输出虚拟化技术相结合的方式来优化网络，默认每个虚拟机都分配有半虚拟化的虚拟网卡，而单根输入输出虚拟化的虚拟设备仅分配给多级图划分后的位于关键路径节点的虚拟机，以达到提高总体网络带宽利用率和降低并行应用总体执行时间的目的。

Description

一种虚拟网络资源分配方法、系统及电子设备 技术领域
本申请属于计算机虚拟化技术领域，特别涉及一种虚拟网络资源分配方法、系统及电子设备。
背景技术
由于近几年来计算机硬件成本的不断减低和单个计算机计算能力的不断加强，现在并行计算已越来越流行并被广泛接受。同时，为了数据中心资源的高效利用，往往采用虚拟化技术以提高数据中心的资源利用率。所谓虚拟化技术就是指能够在一台物理机上运行多台虚拟机技术，并且该虚拟机在用户、应用软件甚至操作系统看来，几乎与物理机没有区别。虽然这能大幅度提高资源利用率，但也由此引出了不同虚拟机间资源分配的问题，因此，如何有效的分配虚拟资源并最大程度的提高资源利用率一直受到广泛的关注和研究。
在当前的并行网络和分布式网络环境中，对网络资源的并行调度已经成为提升网络性能的一个极其重要的技术。但由于现阶段数据中心应用层与底层脱节、应用与网络脱节，这必将会引起网络资源调度的语义鸿沟，而使整个系统的运行缺乏协调性，从而导致网络资源的利用率不高。
发明内容
本申请提供了一种虚拟网络资源分配方法、系统及电子设备，旨在至少在一定程度上解决现有技术中的上述技术问题之一。
为了解决上述问题，本申请提供了如下技术方案：
一种虚拟网络资源分配方法，包括以下步骤：
步骤a：根据物理机上所有虚拟机的网络信息生成任务图，并采用多级图划分算法对所述任务图进行划分，得到k个互不相交的分割子集；
步骤b：计算各个分割子集中具有最长完成时间的关键路径，得到位于关 键路径节点的虚拟机；
步骤c：为所述处于关键路径节点的虚拟机分配单根输入输出虚拟化的虚拟设备。
本申请实施例采取的技术方案还包括：所述步骤a前还包括：收集各个物理机上所有虚拟机的网络信息；所述网络信息的收集方式为：每个虚拟机通过读取UNIX操作系统为用户访问内核提供的伪文件系统获取网络信息，所述虚拟机获取的网络信息放入存储系统中，各个物理机存储系统读取所有虚拟机的网络信息，并将所述网络信息上传至主机。
本申请实施例采取的技术方案还包括：在所述步骤a中，所述根据物理机上所有虚拟机的网络信息生成任务图，并采用多级图划分算法对任务图进行划分具体为：所述主机接收所有物理机上传的网络信息，根据所述网络信息生成带有参数的任务图，并采用多级递归二分划分方法对所述任务图进行划分。
本申请实施例采取的技术方案还包括：所述采用多级递归二分划分方法对任务图进行划分具体为：所述多级递归二分划分方法包括粗化、初始划分和细化三个阶段，所述粗化阶段在输入任务图后产生一系列复杂度更小的聚合图，通过所述聚合图的最大邻接对的顶点降低任务图的复杂度，当所述任务图的复杂度降到预设顶点数量时结束粗化阶段；所述初始划分阶段使用启发式算法对任务图进行初始划分，所述细化阶段将初始划分后的任务图逐步恢复。
本申请实施例采取的技术方案还包括：在所述步骤b中，所述关键路径的计算公式为：
Figure PCTCN2019121297-appb-000001
PathCompTime(CP)≥PathCompTime(P)
上述公式中，path指的是任务图中的某一路径，PathSet指的是任务图中 的全体路径集合，PathCompTime是指路径中所有节点执行完成的时间，符号P是指任意路径，CP是指关键路径。
本申请实施例采取的另一技术方案为：一种虚拟网络资源分配系统，包括物理机和主机，所述主机包括：
多级图划分模块：用于根据物理机上所有虚拟机的网络信息生成任务图，并采用多级图划分算法对所述任务图进行划分，得到k个互不相交的分割子集；
关键路径计算模块：用于计算各个分割子集中具有最长完成时间的关键路径，得到位于关键路径节点的虚拟机；
网络资源分配模块：用于为所述处于关键路径节点的虚拟机分配单根输入输出虚拟化的虚拟设备。
本申请实施例采取的技术方案还包括：所述物理机包括网络资源监控模块，所述网络资源监控模块用于收集物理机上所有虚拟机的网络信息，并将收集的网络信息上传给网络资源管理模块；所述网络信息的收集方式为：每个虚拟机通过读取UNIX操作系统为用户访问内核提供的伪文件系统获取网络信息，所述虚拟机获取的网络信息放入存储系统中，各个物理机存储系统读取所有虚拟机的网络信息，并将所述网络信息上传至网络资源管理模块。
本申请实施例采取的技术方案还包括：所述主机还包括网络资源管理模块，所述网络资源管理模块用于接收所有物理机上传的网络信息，并将网络信息输入多级图划分模块；所述多级图划分模块根据网络信息生成带有参数的任务图，并采用多级递归二分划分方法对所述任务图进行划分。
本申请实施例采取的技术方案还包括：所述采用多级递归二分划分方法对任务图进行划分具体为：所述多级递归二分划分方法包括粗化、初始划分和细化三个阶段，所述粗化阶段在输入任务图后产生一系列复杂度更小的聚合图， 通过所述聚合图的最大邻接对的顶点降低任务图的复杂度，当所述任务图的复杂度降到预设顶点数量时结束粗化阶段；所述初始划分阶段使用启发式算法对任务图进行初始划分，所述细化阶段将初始划分后的任务图逐步恢复。
本申请实施例采取的技术方案还包括：所述关键路径的计算公式为：
Figure PCTCN2019121297-appb-000002
PathCompTime(CP)≥PathCompTime(P)
上述公式中，path指的是任务图中的某一路径，PathSet指的是任务图中的全体路径集合，PathCompTime是指路径中所有节点执行完成的时间，符号P是指任意路径，CP是指关键路径。
本申请实施例采取的又一技术方案为：一种电子设备，包括：
至少一个处理器；以及
与所述至少一个处理器通信连接的存储器；其中，
所述存储器存储有可被所述一个处理器执行的指令，所述指令被所述至少一个处理器执行，以使所述至少一个处理器能够执行上述的虚拟网络资源分配方法的以下操作：
步骤a：根据物理机上所有虚拟机的网络信息生成任务图，并采用多级图划分算法对所述任务图进行划分，得到k个互不相交的分割子集；
步骤b：计算各个分割子集中具有最长完成时间的关键路径，得到位于关键路径节点的虚拟机；
步骤c：为所述处于关键路径节点的虚拟机分配单根输入输出虚拟化的虚拟设备。
相对于现有技术，本申请实施例产生的有益效果在于：本申请实施例的虚拟网络资源分配方法、系统及电子设备采用半虚拟化技术与单根输入输出虚拟 化技术相结合的方式来优化网络，默认每个虚拟机都分配有半虚拟化的虚拟网卡，而单根输入输出虚拟化的虚拟设备仅分配给多级图划分后的位于关键路径节点的虚拟机，以达到提高总体网络带宽利用率和降低并行应用总体执行时间的目的。
附图说明
图1是本申请实施例的虚拟网络资源分配方法的流程图；
图2是本申请实施例的虚拟网络资源分配系统的结构示意图；
图3是本申请实施例的虚拟网络资源分配方法的硬件设备结构示意图。
具体实施方式
为了使本申请的目的、技术方案及优点更加清楚明白，以下结合附图及实施例，对本申请进行进一步详细说明。应当理解，此处所描述的具体实施例仅用以解释本申请，并不用于限定本申请。
针对现有技术存在的问题，本申请实施例的虚拟网络资源分配方法采用半虚拟化技术与单根输入输出虚拟化技术(SR-IOV，Single-Root I/O Virtualization)相结合的方式来优化网络，默认每个虚拟机都分配有半虚拟化的虚拟网卡，而单根输入输出虚拟化的虚拟设备(VF，Virtual Function)仅分配给多级图划分后的位于关键路径(Critical Path)节点的虚拟机，以达到提高总体网络带宽利用率和降低并行应用总体执行时间的目的。
具体地，请参阅图1，是本申请实施例的虚拟网络资源分配方法的流程图。本申请实施例的虚拟网络资源分配方法包括以下步骤：
步骤100：收集各个物理机上所有虚拟机的网络信息，并将收集的网络信息上传给主机；
步骤100中，网络信息的收集方式为：每个虚拟机通过读取UNIX操作系统为用户访问内核提供的伪文件系统/proc/net/dev来获取网络信息，虚拟机获取的网络信息放入存储系统(XenStore)中，各个物理机上分别设有网络资源监控模块，网络资源监控模块通过XenStore读取各个虚拟机的网络信息，并发送给网络资源管理模块。
步骤200：通过主机接收各个物理机上所有虚拟机的网络信息，生成带有参数的任务图，并采用多级图划分算法对任务图进行划分操作，得到k个互不相交的分割子集；
步骤200中，主机上包括网络资源管理模块和多级图划分模块，网络资源管理模块负责接收各个物理机上所有虚拟机的网络信息，并将网络信息输入多级图划分模块，生成带有参数的任务图，多级图划分模块采用多级递归二分划分方法对任务图进行划分，得到k个互不相交的分割子集，不同分割子集对应的任务分配给不同的物理机进行计算。本申请通过对任务图进行多级图划分，从而平衡各个物理机之间的计算量，并使各个物理机之间的总通信量最小。
具体地，多级递归二分划分方法包括粗化、初始划分和细化三个阶段。粗化阶段在输入任务图后将产生一系列复杂度更小的聚合图，目的在于通过聚合图的最大邻接对的顶点以降低任务图的复杂度，当任务图的复杂度降到预设顶点数量(通常顶点数量为几百个，具体顶点数量可根据实际应用进行设定)时结束粗化阶段。在初始划分阶段，因为任务图的复杂度已经很小了，使用经典简单的启发式算法如K-L算法对任务图进行初始划分。在细化阶段，将初始划分后的任务图逐步恢复，将聚合在一起的顶点在不同的分割子集间移动，提高任务图的划分效果。
步骤300：计算各个分割子集中具有最长完成时间的关键路径，得到各个物理机上位于关键路径节点的虚拟机；
步骤300中，由于在同一物理机间的虚拟机通信开销可视为零，因此在任务图划分后的关键路径计算只计算不同分割子集间的通信开销。具体地，关键路径的计算公式为：
Figure PCTCN2019121297-appb-000003
PathCompTime(CP)≥PathCompTime(P)  (1)
公式(1)中，path指的是任务图中的某一路径，PathSet指的是任务图中的全体路径集合，PathCompTime是指路径中所有节点执行完成的时间，符号P是指任意路径，CP是指关键路径。
步骤400：为各个物理机上处于关键路径节点的虚拟机分配单根输入输出虚拟化的虚拟设备；
步骤400中，本申请实施例的网络资源分配方法是默认让每个虚拟机都分配有半虚拟化的虚拟网卡，而单根输入输出虚拟化的虚拟设备仅分配给多级图划分后的位于关键路径节点的虚拟机。其中半虚拟化的虚拟网卡与单根输入输出虚拟化的虚拟设备之间透明切换使用了多网卡绑定技术，单根输入输出虚拟化的虚拟设备被设置为主网卡，半虚拟化的虚拟网卡被设置为从网卡，当主网卡有效的时候虚拟机使用主网卡进行通信，只有在主网卡无效的时候才使用从网卡进行通信，从而减少网络开销。
步骤500：判定所有分割子集是否执行完毕，如果所有分割子集执行完毕，执行步骤600；否则，返回步骤300继续计算下一分割子集的关键路径计算，直到所有分割子集执行完毕；
步骤600：虚拟网络资源分配结束。
请参阅图2，是本申请实施例的虚拟网络资源分配系统的结构图。本申请实施例的虚拟网络资源分配系统包括物理机和主机，物理机包括网络资源监控模块，主机包括网络资源管理模块、多级图划分模块、关键路径计算模块、网络资源分配模块和任务执行判断模块。
网络资源监控模块：用于收集各个物理机上所有虚拟机的网络信息，并将收集的网络信息上传给网络资源管理模块；其中，网络信息的收集方式为：每个虚拟机通过读取UNIX操作系统为用户访问内核提供的伪文件系统/proc/net/dev来获取网络信息，虚拟机获取的网络信息放入XenStore中，网络资源监控模块通过XenStore读取各个虚拟机的网络信息，并发送给网络资源管理模块。
网络资源管理模块：用于接收各个物理机上所有虚拟机的网络信息，并将网络信息输入多级图划分模块；
多级图划分模块：用于根据输入的网络信息生成带有参数的任务图，并采用多级图划分算法对任务图进行划分操作，得到k个互不相交的分割子集；其中，多级图划分模块采用多级递归二分划分方法对任务图进行划分，得到k个互不相交的分割子集，不同分割子集对应的任务分配给不同的物理机进行计算。本申请通过对任务图进行多级图划分，从而平衡各个物理机之间的计算量，并使各个物理机之间的总通信量最小。
具体地，多级递归二分划分方法包括粗化、初始划分和细化三个阶段。粗化阶段在输入任务图后将产生一系列复杂度更小的聚合图，目的在于通过聚合图的最大邻接对的顶点以降低任务图的复杂度，当任务图的复杂度降到只有几百个顶点时结束粗化阶段。在初始划分阶段，因为任务图的复杂度已经很小了，使用经典简单的启发式算法如K-L算法对任务图进行初始划分。在细化阶段， 将初始划分后的任务图逐步恢复，将聚合在一起的顶点在不同的分割子集间移动，提高任务图的划分效果。
关键路径计算模块：用于计算各个分割子集中具有最长完成时间的关键路径，得到各个物理机上位于关键路径节点的虚拟机；其中，由于在同一物理机间的虚拟机通信开销可视为零，因此在任务图划分后的关键路径计算只计算不同分割子集间的通信开销。具体地，关键路径的计算公式为：
Figure PCTCN2019121297-appb-000004
PathCompTime(CP)≥PathCompTime(P)  (1)
公式(1)中，path指的是任务图中的某一路径，PathSet指的是任务图中的全体路径集合，PathCompTime是指路径中所有节点执行完成的时间，符号P是指任意路径，CP是指关键路径。
网络资源分配模块：用于为各个物理机上处于关键路径节点的虚拟机分配单根输入输出虚拟化的虚拟设备；其中，本申请实施例的网络资源分配方法是默认让每个虚拟机都分配有半虚拟化的虚拟网卡，而单根输入输出虚拟化的虚拟设备仅分配给多级图划分后的位于关键路径节点的虚拟机。其中半虚拟化的虚拟网卡与单根输入输出虚拟化的虚拟设备之间透明切换使用了多网卡绑定技术，单根输入输出虚拟化的虚拟设备被设置为主网卡，半虚拟化的虚拟网卡被设置为从网卡，当主网卡有效的时候虚拟机使用主网卡进行通信，只有在主网卡无效的时候才使用从网卡进行通信，从而减少网络开销。
任务执行判断模块：用于判定所有分割子集是否执行完毕，如果所有分割子集执行完毕，则虚拟网络资源分配结束；否则，通过关键路径计算模块继续计算下一分割子集的关键路径计算，直到所有分割子集执行完毕。
图3是本申请实施例提供的虚拟网络资源分配方法的硬件设备结构示意图。 如图3所示，该设备包括一个或多个处理器以及存储器。以一个处理器为例，该设备还可以包括：输入系统和输出系统。
处理器、存储器、输入系统和输出系统可以通过总线或者其他方式连接，图3中以通过总线连接为例。
存储器作为一种非暂态计算机可读存储介质，可用于存储非暂态软件程序、非暂态计算机可执行程序以及模块。处理器通过运行存储在存储器中的非暂态软件程序、指令以及模块，从而执行电子设备的各种功能应用以及数据处理，即实现上述方法实施例的处理方法。
存储器可以包括存储程序区和存储数据区，其中，存储程序区可存储操作系统、至少一个功能所需要的应用程序；存储数据区可存储数据等。此外，存储器可以包括高速随机存取存储器，还可以包括非暂态存储器，例如至少一个磁盘存储器件、闪存器件、或其他非暂态固态存储器件。在一些实施例中，存储器可选包括相对于处理器远程设置的存储器，这些远程存储器可以通过网络连接至处理系统。上述网络的实例包括但不限于互联网、企业内部网、局域网、移动通信网及其组合。
输入系统可接收输入的数字或字符信息，以及产生信号输入。输出系统可包括显示屏等显示设备。
所述一个或者多个模块存储在所述存储器中，当被所述一个或者多个处理器执行时，执行上述任一方法实施例的以下操作：
步骤a：根据物理机上所有虚拟机的网络信息生成任务图，并采用多级图划分算法对所述任务图进行划分，得到k个互不相交的分割子集；
步骤b：计算各个分割子集中具有最长完成时间的关键路径，得到位于关键路径节点的虚拟机；
步骤c：为所述处于关键路径节点的虚拟机分配单根输入输出虚拟化的虚拟设备。
上述产品可执行本申请实施例所提供的方法，具备执行方法相应的功能模块和有益效果。未在本实施例中详尽描述的技术细节，可参见本申请实施例提供的方法。
本申请实施例提供了一种非暂态(非易失性)计算机存储介质，所述计算机存储介质存储有计算机可执行指令，该计算机可执行指令可执行以下操作：
步骤a：根据物理机上所有虚拟机的网络信息生成任务图，并采用多级图划分算法对所述任务图进行划分，得到k个互不相交的分割子集；
步骤b：计算各个分割子集中具有最长完成时间的关键路径，得到位于关键路径节点的虚拟机；
步骤c：为所述处于关键路径节点的虚拟机分配单根输入输出虚拟化的虚拟设备。
本申请实施例提供了一种计算机程序产品，所述计算机程序产品包括存储在非暂态计算机可读存储介质上的计算机程序，所述计算机程序包括程序指令，当所述程序指令被计算机执行时，使所述计算机执行以下操作：
步骤a：根据物理机上所有虚拟机的网络信息生成任务图，并采用多级图划分算法对所述任务图进行划分，得到k个互不相交的分割子集；
步骤b：计算各个分割子集中具有最长完成时间的关键路径，得到位于关键路径节点的虚拟机；
步骤c：为所述处于关键路径节点的虚拟机分配单根输入输出虚拟化的虚拟设备。
本申请实施例的虚拟网络资源分配方法、系统及电子设备采用半虚拟化技术与单根输入输出虚拟化技术相结合的方式来优化网络，默认每个虚拟机都分配有半虚拟化的虚拟网卡，而单根输入输出虚拟化的虚拟设备仅分配给多级图 划分后的位于关键路径节点的虚拟机，以达到提高总体网络带宽利用率和降低并行应用总体执行时间的目的。
以上所述仅是本发明的优选实施方式，应当指出，对于本技术领域的普通技术人员来说，在不脱离本发明原理的前提下，还可以做出若干改进和润饰，这些改进和润饰也应视为本发明的保护范围。

Claims (11)

  1. 一种虚拟网络资源分配方法，其特征在于，包括以下步骤：
    步骤a：根据物理机上所有虚拟机的网络信息生成任务图，并采用多级图划分算法对所述任务图进行划分，得到k个互不相交的分割子集；
    步骤b：计算各个分割子集中具有最长完成时间的关键路径，得到位于关键路径节点的虚拟机；
    步骤c：为所述处于关键路径节点的虚拟机分配单根输入输出虚拟化的虚拟设备。
  2. 根据权利要求1所述的虚拟网络资源分配方法，其特征在于，所述步骤a前还包括：收集各个物理机上所有虚拟机的网络信息；所述网络信息的收集方式为：每个虚拟机通过读取UNIX操作系统为用户访问内核提供的伪文件系统获取网络信息，所述虚拟机获取的网络信息放入存储系统中，各个物理机存储系统读取所有虚拟机的网络信息，并将所述网络信息上传至主机。
  3. 根据权利要求2所述的虚拟网络资源分配方法，其特征在于，在所述步骤a中，所述根据物理机上所有虚拟机的网络信息生成任务图，并采用多级图划分算法对任务图进行划分具体为：所述主机接收所有物理机上传的网络信息，根据所述网络信息生成带有参数的任务图，并采用多级递归二分划分方法对所述任务图进行划分。
  4. 根据权利要求3所述的虚拟网络资源分配方法，其特征在于，所述采用多级递归二分划分方法对任务图进行划分具体为：所述多级递归二分划分方法包括粗化、初始划分和细化三个阶段，所述粗化阶段在输入任务图后产生一系列复杂度更小的聚合图，通过所述聚合图的最大邻接对的顶点降低任务图的复 杂度，当所述任务图的复杂度降到预设顶点数量时结束粗化阶段；所述初始划分阶段使用启发式算法对任务图进行初始划分，所述细化阶段将初始划分后的任务图逐步恢复。
  5. 根据权利要求1至4任一项所述的虚拟网络资源分配方法，其特征在于，在所述步骤b中，所述关键路径的计算公式为：
    Figure PCTCN2019121297-appb-100001
    PathCompTime(CP)≥PathCompTime(P)
    上述公式中，path指的是任务图中的某一路径，PathSet指的是任务图中的全体路径集合，PathCompTime是指路径中所有节点执行完成的时间，符号P是指任意路径，CP是指关键路径。
  6. 一种虚拟网络资源分配系统，包括物理机和主机，其特征在于，所述主机包括：
    多级图划分模块：用于根据物理机上所有虚拟机的网络信息生成任务图，并采用多级图划分算法对所述任务图进行划分，得到k个互不相交的分割子集；
    关键路径计算模块：用于计算各个分割子集中具有最长完成时间的关键路径，得到位于关键路径节点的虚拟机；
    网络资源分配模块：用于为所述处于关键路径节点的虚拟机分配单根输入输出虚拟化的虚拟设备。
  7. 根据权利要求6所述的虚拟网络资源分配系统，其特征在于，所述物理机包括网络资源监控模块，所述网络资源监控模块用于收集物理机上所有虚拟机的网络信息，并将收集的网络信息上传给网络资源管理模块；所述网络信息的收集方式为：每个虚拟机通过读取UNIX操作系统为用户访问内核提供的伪文件系统获取网络信息，所述虚拟机获取的网络信息放入存储系统中，各个物理机存储系统读取所有虚拟机的网络信息，并将所述网络信息上传至网络资源管理模块。
  8. 根据权利要求7所述的虚拟网络资源分配系统，其特征在于，所述主机还包括网络资源管理模块，所述网络资源管理模块用于接收所有物理机上传的网络信息，并将网络信息输入多级图划分模块；所述多级图划分模块根据网络信息生成带有参数的任务图，并采用多级递归二分划分方法对所述任务图进行划分。
  9. 根据权利要求8所述的虚拟网络资源分配系统，其特征在于，所述采用多级递归二分划分方法对任务图进行划分具体为：所述多级递归二分划分方法包括粗化、初始划分和细化三个阶段，所述粗化阶段在输入任务图后产生一系列复杂度更小的聚合图，通过所述聚合图的最大邻接对的顶点降低任务图的复杂度，当所述任务图的复杂度降到预设顶点数量时结束粗化阶段；所述初始划分阶段使用启发式算法对任务图进行初始划分，所述细化阶段将初始划分后的任务图逐步恢复。
  10. 根据权利要求6至9任一项所述的虚拟网络资源分配系统，其特征在于，所述关键路径的计算公式为：
    Figure PCTCN2019121297-appb-100002
    PathCompTime(CP)≥PathCompTime(P)
    上述公式中，path指的是任务图中的某一路径，PathSet指的是任务图中的全体路径集合，PathCompTime是指路径中所有节点执行完成的时间，符号P是指任意路径，CP是指关键路径。
  11. 一种电子设备，包括：
    至少一个处理器；以及
    与所述至少一个处理器通信连接的存储器；其中，
    所述存储器存储有可被所述一个处理器执行的指令，所述指令被所述至少一个处理器执行，以使所述至少一个处理器能够执行上述1至5任一项所述的 虚拟网络资源分配方法的以下操作：
    步骤a：根据物理机上所有虚拟机的网络信息生成任务图，并采用多级图划分算法对所述任务图进行划分，得到k个互不相交的分割子集；
    步骤b：计算各个分割子集中具有最长完成时间的关键路径，得到位于关键路径节点的虚拟机；
    步骤c：为所述处于关键路径节点的虚拟机分配单根输入输出虚拟化的虚拟设备。
PCT/CN2019/121297 2018-11-28 2019-11-27 一种虚拟网络资源分配方法、系统及电子设备 Ceased WO2020108536A1 (zh)

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN201811430362.9 2018-11-28
CN201811430362.9A CN109412865B (zh) 2018-11-28 2018-11-28 一种虚拟网络资源分配方法、系统及电子设备

Publications (1)

Publication Number Publication Date
WO2020108536A1 true WO2020108536A1 (zh) 2020-06-04

Family

ID=65455905

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2019/121297 Ceased WO2020108536A1 (zh) 2018-11-28 2019-11-27 一种虚拟网络资源分配方法、系统及电子设备

Country Status (2)

Country Link
CN (1) CN109412865B (zh)
WO (1) WO2020108536A1 (zh)

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN112330229A (zh) * 2020-12-02 2021-02-05 北京元心科技有限公司 资源调度方法、装置、电子设备及计算机可读存储介质
CN121441683A (zh) * 2025-12-31 2026-01-30 杭州景唐通信技术有限公司 基于异构计算及动态能效的网关处理方法及装置

Families Citing this family (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN109412865B (zh) * 2018-11-28 2020-08-25 深圳先进技术研究院 一种虚拟网络资源分配方法、系统及电子设备
CN111427912A (zh) * 2020-03-31 2020-07-17 拉卡拉支付股份有限公司 一种任务处理方法、装置、电子设备及存储介质

Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN105357322A (zh) * 2015-12-11 2016-02-24 中国科学院信息工程研究所 一种基于拓扑划分的虚拟机分配方法
CN108205461A (zh) * 2016-12-19 2018-06-26 华耀(中国)科技有限公司 一种混合部署的虚拟化平台及部署方法
CN109412865A (zh) * 2018-11-28 2019-03-01 深圳先进技术研究院 一种虚拟网络资源分配方法、系统及电子设备

Family Cites Families (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US9923837B2 (en) * 2013-08-29 2018-03-20 Ericsson Ab Method and system to allocate bandwidth based on task deadline in cloud computing networks
US10261813B2 (en) * 2013-09-25 2019-04-16 Arm Limited Data processing system for dispatching tasks from a plurality of applications to a shared resource provided by an accelerator
CN104536806B (zh) * 2014-12-26 2017-11-03 东南大学 一种云环境下的工作流应用弹性资源供应方法
CN110113184B (zh) * 2019-04-17 2021-10-26 杭州中科先进技术研究院有限公司 Sr-iov环境下的kvm虚拟机网络优化方法及装置

Patent Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN105357322A (zh) * 2015-12-11 2016-02-24 中国科学院信息工程研究所 一种基于拓扑划分的虚拟机分配方法
CN108205461A (zh) * 2016-12-19 2018-06-26 华耀(中国)科技有限公司 一种混合部署的虚拟化平台及部署方法
CN109412865A (zh) * 2018-11-28 2019-03-01 深圳先进技术研究院 一种虚拟网络资源分配方法、系统及电子设备

Cited By (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN112330229A (zh) * 2020-12-02 2021-02-05 北京元心科技有限公司 资源调度方法、装置、电子设备及计算机可读存储介质
CN112330229B (zh) * 2020-12-02 2023-09-22 北京元心科技有限公司 资源调度方法、装置、电子设备及计算机可读存储介质
CN121441683A (zh) * 2025-12-31 2026-01-30 杭州景唐通信技术有限公司 基于异构计算及动态能效的网关处理方法及装置

Also Published As

Publication number Publication date
CN109412865A (zh) 2019-03-01
CN109412865B (zh) 2020-08-25

Similar Documents

Publication Publication Date Title
US9563458B2 (en) Offloading and parallelizing translation table operations
US20200065208A1 (en) Model-based key performance indicator service for data analytics processing platforms
US20200272526A1 (en) Methods and systems for automated scaling of computing clusters
CN110554917A (zh) 高效遍历较大数据量集合的方法、系统、终端及存储介质
CN103455363A (zh) 一种虚拟机的指令处理方法、装置及物理主机
US9755986B1 (en) Techniques for tightly-integrating an enterprise storage array into a distributed virtualized computing environment
WO2020108536A1 (zh) 一种虚拟网络资源分配方法、系统及电子设备
US20240012683A1 (en) Resource Management Unit for Capturing Operating System Configuration States and Offloading Tasks
CN114281516B (zh) 一种基于numa属性的资源分配方法及装置
CN106325756A (zh) 一种数据存储、数据计算方法和设备
WO2017000645A1 (zh) 一种分配宿主机资源的方法和装置
US11630698B2 (en) Resource management unit for capturing operating system configuration states and swapping memory content
CN107562520A (zh) 扩容虚拟机的内存的方法和装置
EP4042305B1 (en) Resource management unit for capturing operating system configuration states and managing malware
CN109614242B (zh) 一种计算能力共享方法、装置、设备及介质
CN117076720B (zh) 一种嵌入表访问方法、装置、电子设备及存储介质
US20150199217A1 (en) Entropy resource allocation management in virtualized environments
US20230012021A1 (en) Feature Resource Self-Tuning and Rebalancing
US11481255B2 (en) Management of memory pages for a set of non-consecutive work elements in work queue designated by a sliding window for execution on a coherent accelerator
CN111427887A (zh) 一种快速扫描HBase分区表的方法、装置、系统
CN112506582B (zh) 小程序数据包处理方法、装置、设备及介质
CN115309502A (zh) 一种容器调度方法及装置
CN119536987A (zh) 一种容器资源管理方法、装置、电子设备及介质
CN104793985B (zh) 一种虚拟机调度方法和管理设备
CN116560803A (zh) 基于sr-iov的资源管理方法及相关装置

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 19888403

Country of ref document: EP

Kind code of ref document: A1

NENP Non-entry into the national phase

Ref country code: DE

32PN Ep: public notification in the ep bulletin as address of the adressee cannot be established

Free format text: NOTING OF LOSS OF RIGHTS PURSUANT TO RULE 112(1) EPC (EPO FORM 1205A DATED 03/11/2021)

122 Ep: pct application non-entry in european phase

Ref document number: 19888403

Country of ref document: EP

Kind code of ref document: A1