CN116974762A - Method and system for scheduling and managing server resources - Google Patents

Method and system for scheduling and managing server resources Download PDF

Info

Publication number
CN116974762A
CN116974762A CN202310910792.5A CN202310910792A CN116974762A CN 116974762 A CN116974762 A CN 116974762A CN 202310910792 A CN202310910792 A CN 202310910792A CN 116974762 A CN116974762 A CN 116974762A
Authority
CN
China
Prior art keywords
virtual machine
resource pool
server resource
hardware
resources
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.)
Pending
Application number
CN202310910792.5A
Other languages
Chinese (zh)
Inventor
邓艳山
袁振涛
蔡财义
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.)
Fiberhome Supermicro Information And Technology Co ltd
Original Assignee
Fiberhome Supermicro Information And Technology Co ltd
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 Fiberhome Supermicro Information And Technology Co ltd filed Critical Fiberhome Supermicro Information And Technology Co ltd
Priority to CN202310910792.5A priority Critical patent/CN116974762A/en
Publication of CN116974762A publication Critical patent/CN116974762A/en
Pending legal-status Critical Current

Links

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F9/00—Arrangements for program control, e.g. control units
    • G06F9/06—Arrangements for program control, e.g. control units using stored programs, i.e. using an internal store of processing equipment to receive or retain programs
    • G06F9/46—Multiprogramming arrangements
    • G06F9/50—Allocation of resources, e.g. of the central processing unit [CPU]
    • G06F9/5061—Partitioning or combining of resources
    • G06F9/5077—Logical partitioning of resources; Management or configuration of virtualized resources
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F9/00—Arrangements for program control, e.g. control units
    • G06F9/06—Arrangements for program control, e.g. control units using stored programs, i.e. using an internal store of processing equipment to receive or retain programs
    • G06F9/44—Arrangements for executing specific programs
    • G06F9/455—Emulation; Interpretation; Software simulation, e.g. virtualisation or emulation of application or operating system execution engines
    • G06F9/45533—Hypervisors; Virtual machine monitors
    • G06F9/45558—Hypervisor-specific management and integration aspects
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F9/00—Arrangements for program control, e.g. control units
    • G06F9/06—Arrangements for program control, e.g. control units using stored programs, i.e. using an internal store of processing equipment to receive or retain programs
    • G06F9/44—Arrangements for executing specific programs
    • G06F9/455—Emulation; Interpretation; Software simulation, e.g. virtualisation or emulation of application or operating system execution engines
    • G06F9/45533—Hypervisors; Virtual machine monitors
    • G06F9/45558—Hypervisor-specific management and integration aspects
    • G06F2009/4557—Distribution of virtual machine instances; Migration and load balancing

Landscapes

  • Engineering & Computer Science (AREA)
  • Software Systems (AREA)
  • Theoretical Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • General Engineering & Computer Science (AREA)
  • General Physics & Mathematics (AREA)
  • Hardware Redundancy (AREA)

Abstract

The application discloses a server resource scheduling management method and a system, which allocate a virtual machine to be deployed corresponding to a service to a corresponding target server resource pool for deployment according to the instruction set type and the hardware type of the service requirement; if the hardware resources of the target server resource pool cannot meet the resource requirements of the virtual machines to be deployed, migrating the virtual machines in the target server resource pool to the heterogeneous server resource pool for deployment and/or restarting the virtual machines in the target server resource pool so as to release the hardware resources. The method and the device realize the perception of a service special instruction set and special hardware requirements, can allocate a proper resource pool for the service to deploy the virtual machine according to the service requirements, realize the unified scheduling and management of resources in the heterogeneous server resource pool, improve the utilization rate of the server resources, realize the balanced load of the heterogeneous server and improve the success rate of the deployment of the virtual machine.

Description

Method and system for scheduling and managing server resources
Technical Field
The application relates to the technical field of cloud computing, in particular to a server resource scheduling management method and system.
Background
In the current computing power network architecture, computing power resources are mainly deployed in a data center server, deployed in a cloud platform mode, and various resources such as a CPU (Central processing Unit), a GPU (graphics processing Unit) accelerator card, an AI (advanced technology interface) accelerator card and the like in the server are managed and scheduled through the cloud platform. Because of the heterogeneous difference of the CPUs, the cloud platform deployed by the server of the X86 architecture and the cloud platform deployed by the server of the ARM architecture are respectively and independently deployed and managed. When the cloud platform manages server resources, the cloud platform of the X86 architecture can only schedule and manage computing power resources of the X86 server, and can only deploy and migrate virtual machines on the computing power resources of the X86 architecture server. The cloud platform of the ARM architecture can only schedule and manage the computing power resources of the ARM server, and can only deploy and migrate virtual machines on the computing power resources of the ARM architecture server.
In the technical context of a computing power network, some services have specific requirements on computing power resources. For example, a business requires a special instruction set, a specific heterogeneous accelerated computing power board resource, or a specific memory resource. Because of the difference of hardware and service load, the X86 resource pool and the ARM resource pool have unbalanced phenomena in the aspects of CPU instruction sets, core numbers, hardware computing resources, computing resource loads and the like. In the prior art, when the computing power resource of the server is insufficient, the capacity of the server is expanded or hardware purchase is required. Under the background of the technical trend of the computing power network, the management of the heterogeneous computing power resource cloud platform at present does not meet the requirements of the development of the computing power network and the improvement of the actual service deployment efficiency.
Therefore, the resources and the load in the server resource pool can only be scheduled inside the server, so that the heterogeneous server load imbalance is a technical problem to be solved.
Disclosure of Invention
The application mainly aims to provide a server resource scheduling management method and system, and aims to solve the technical problem that in the related art, resources and loads in a server resource pool can only be scheduled in a server, so that the server load is unbalanced.
In a first aspect, the present application provides a method for scheduling and managing server resources, the method including the steps of:
distributing the virtual machines to be deployed corresponding to the service to a corresponding target server resource pool for deployment according to the instruction set type and the hardware type of the service requirement;
if the hardware resources of the target server resource pool cannot meet the resource requirements of the virtual machines to be deployed, migrating the virtual machines in the target server resource pool to the heterogeneous server resource pool for deployment and/or restarting the virtual machines in the target server resource pool so as to release the hardware resources.
In some embodiments, if the hardware resources of the target server resource pool cannot meet the resource requirements of the virtual machine to be deployed, migrating the virtual machine in the target server resource pool to the heterogeneous server resource pool for deployment and/or restarting the virtual machine in the target server resource pool to release the hardware resources, including:
If the conventional hardware resources of the target server resource pool cannot meet the conventional hardware resource requirements of the virtual machine to be deployed, migrating the pure operation virtual machine in the target server resource pool to a heterogeneous server resource pool heterogeneous with the target server resource pool so as to release the conventional hardware resources;
restarting the virtual machine with idle specific hardware resources in the target server resource pool to release the specific hardware resources if the specific hardware resources in the target server resource pool are insufficient;
the conventional hardware resources comprise CPU resources, memory resources and disk resources, and the specific hardware resources comprise AI acceleration card resources and GPU acceleration card resources;
the pure operation virtual machine is a virtual machine without specific type of hardware requirements.
In some embodiments, if the conventional hardware resources of the target server resource pool cannot meet the conventional hardware resource requirements of the virtual machine to be deployed, migrating the pure operation virtual machine in the target server resource pool to a heterogeneous server resource pool heterogeneous with the target server resource pool to release the conventional hardware resources, including:
according to the demand of the virtual machine to be deployed for each conventional hardware resource, determining a first conventional hardware resource which cannot meet the demand of the virtual machine to be deployed in a target server resource pool;
Determining the occupation amount of each pure operation virtual machine in a target server resource pool to each conventional hardware resource;
multiplying the occupation amount of each pure operation virtual machine to each conventional hardware resource by a weight coefficient corresponding to each conventional hardware resource, and adding the results to obtain a migration coefficient of each pure operation virtual machine, wherein the weight coefficient of the first conventional hardware resource is the largest;
determining whether a pure operation virtual machine with the occupation amount of the first conventional hardware resources being larger than or equal to the first conventional demand amount of the virtual machine to be deployed for the first conventional hardware resources exists;
if yes, the pure operation virtual machine with the first conventional hardware resource occupation amount being larger than or equal to the first conventional demand amount and the minimum migration coefficient is used as a migration virtual machine and migrated to the heterogeneous server resource pool;
otherwise, the plurality of pure operation virtual machines with the minimum sum of migration coefficients are taken as migration virtual machines and migrated to the heterogeneous server resource pool, wherein the sum of the first conventional hardware resource occupation amount is larger than or equal to the first conventional demand amount.
In some embodiments, migrating the migration virtual machine into the heterogeneous server resource pool includes:
Inquiring the instruction set type, format and disk size of the migration virtual machine;
determining whether the instruction set is a specific type instruction set corresponding to the target server resource pool according to the instruction set type of the migration virtual machine;
if yes, canceling migration of the migration virtual machine, and returning migration failure information;
otherwise, creating an empty mirror image file in the heterogeneous server resource pool, and configuring a new virtual machine according to the format and the disk size of the virtual machine to be migrated;
and translating the format of the instruction set of the virtual machine to be migrated into the format of the instruction set of the heterogeneous server resource pool, and writing the translated instruction set into the empty mirror image file.
In some embodiments, the method further comprises:
determining the occupancy rate of each specific hardware resource in each current resource pool according to the total number of PCIE cards in each specific hardware of each server resource pool, the set maximum power consumption of the PCIE cards and the actual total power consumption of the current PCIE cards;
and if the occupancy rate of the specific hardware resources is larger than or equal to a preset occupancy rate threshold value, sending out a warning of insufficient specific hardware resources.
In some embodiments, if the specific hardware resources in the target server resource pool are insufficient, restarting the virtual machine with the free specific hardware resources in the target server resource pool to release the specific hardware resources includes:
Determining the idle quantity of the first specific hardware resources in each virtual machine according to the occupation quantity of the first specific hardware resources with insufficient resources of each virtual machine in the target resource pool, the real-time power consumption of the PCIE card in the first specific hardware and the set maximum power consumption of the PCIE card;
determining whether there is a virtual machine having an amount of idleness for the first specific hardware resource greater than or equal to a first specific demand for the first specific hardware resource by the virtual machine to be deployed;
if yes, taking the virtual machine with the largest first specific hardware resource idle amount and larger than or equal to the first specific demand as a virtual machine to be restarted, configuring the occupation amount of the first specific hardware resource of the virtual machine to be restarted as the occupation amount of the first specific hardware resource before restarting minus the first specific demand, and restarting the virtual machine to be restarted to release the first specific hardware resource;
otherwise, taking the plurality of virtual machines with the sum of the idle amounts of the first specific hardware resources being greater than or equal to the first specific demand as virtual machines to be restarted, configuring the total occupied amount of the first specific hardware resources of the plurality of virtual machines to be restarted as the total occupied amount of the first specific hardware resources before the virtual machines are restarted minus the first specific demand, and restarting the virtual machines to be restarted to release the first specific hardware resources.
In some embodiments, the allocating the virtual machine to be deployed corresponding to the service to the corresponding target server resource pool for deployment according to the instruction set type and the hardware type of the service requirement includes:
if the service requires a specific type instruction set, distributing the virtual machine to be deployed to a target server resource pool of the specific type instruction set for deployment;
if the service requires the specific type of hardware, the virtual machine to be deployed is distributed to a target server resource pool containing the specific type of hardware for deployment;
if the service requires a specific type instruction set and specific type hardware, judging whether a target server resource pool of the specific type instruction set comprises the specific type hardware, if so, distributing the virtual machine to be deployed to the target server resource pool for deployment, otherwise, marking that the virtual machine to be deployed cannot be deployed, and sending alarm information of the lack of the specific type hardware in the target server resource pool;
and if the service does not have a specific type instruction set and specific type hardware requirements, distributing the virtual machine to be deployed to a target server resource pool corresponding to the mirror image type of the virtual machine to be deployed.
In some embodiments, the server resource pools of the different instruction sets are heterogeneous server resource pools with respect to each other.
In a second aspect, the present application further provides a server resource scheduling management system, where the system includes:
the allocation module is used for allocating the virtual machine to be deployed corresponding to the service to a corresponding target server resource pool for deployment according to the instruction set type and the hardware type of the service requirement;
and the resource management module is used for migrating the virtual machines in the target server resource pool to the heterogeneous server resource pool for deployment and/or restarting the virtual machines in the target server resource pool to release the hardware resources if the hardware resources of the target server resource pool cannot meet the resource requirements of the virtual machines to be deployed.
In some embodiments, the server resource scheduling management system is disposed in a cloud platform corresponding to the target server resource pool and a server at a higher level of the cloud platform corresponding to the heterogeneous resource pool.
The application provides a server resource scheduling management method and a system, which allocate a virtual machine to be deployed corresponding to a service to a corresponding target server resource pool for deployment according to the instruction set type and the hardware type of the service requirement; if the hardware resources of the target server resource pool cannot meet the resource requirements of the virtual machines to be deployed, migrating the virtual machines in the target server resource pool to the heterogeneous server resource pool for deployment and/or restarting the virtual machines in the target server resource pool so as to release the hardware resources. The method and the device realize the perception of a service special instruction set and special hardware requirements, can allocate a proper resource pool for the service to deploy the virtual machine according to the service requirements, realize the unified scheduling and management of resources in the heterogeneous server resource pool, improve the utilization rate of the server resources, realize the balanced load of the heterogeneous server and improve the success rate of the deployment of the virtual machine.
Drawings
In order to more clearly illustrate the technical solutions of the embodiments of the present application, the drawings required for the description of the embodiments will be briefly described below, and it is obvious that the drawings in the following description are some embodiments of the present application, and other drawings may be obtained according to these drawings without inventive effort for a person skilled in the art.
Fig. 1 is a schematic flow chart of a server resource scheduling management method according to an embodiment of the present application;
FIG. 2 is a schematic diagram of a set position of a server resource scheduling management system;
FIG. 3 is a schematic diagram of a specific flow of deploying virtual machines in an X86 server resource pool;
FIG. 4 is a schematic diagram of a specific flow of deploying virtual machines in an ARM server resource pool;
fig. 5 is a schematic block diagram of a server resource scheduling management system according to an embodiment of the present application.
The achievement of the objects, functional features and advantages of the present application will be further described with reference to the accompanying drawings, in conjunction with the embodiments.
Detailed Description
The following description of the embodiments of the present application will be made clearly and fully with reference to the accompanying drawings, in which it is evident that the embodiments described are some, but not all embodiments of the application. All other embodiments, which can be made by those skilled in the art based on the embodiments of the application without making any inventive effort, are intended to be within the scope of the application.
The flow diagrams depicted in the figures are merely illustrative and not necessarily all of the elements and operations/steps are included or performed in the order described. For example, some operations/steps may be further divided, combined, or partially combined, so that the order of actual execution may be changed according to actual situations.
The embodiment of the application provides a server resource scheduling management method and system.
Some embodiments of the present application are described in detail below with reference to the accompanying drawings. The following embodiments and features of the embodiments may be combined with each other without conflict.
It is worth to say that in the current cloud platform server computing power resource management scheme, the cloud platform is separately deployed by the X86 server computing power resource and the ARM server computing power resource, the X86 cloud platform and the ARM cloud platform respectively manage the corresponding computing power resource and perform virtual machine service deployment, computing power resources of a heterogeneous server resource pool cannot be scheduled, meanwhile, the cloud platform does not perceive the instruction set requirement and the hardware requirement of the service, only creates a virtual machine in the computing power resource pool managed by the corresponding cloud platform according to the virtual machine configuration information set by a user, and when the resource pool does not have idle resources, a result of failure in creating the virtual machine is returned, or the problem that the virtual machine is down or performance does not meet the service requirement due to the fact that the instruction set or the hardware computing power resource is not satisfied is solved. In the related art, successful deployment of the virtual machine is generally ensured by expanding a server or adding hardware computing power resources, which greatly increases hardware purchasing cost of the computing power resources and can influence deployment time and service online time of the virtual machine. And the imbalance of hardware resources and service loads of the X86 and ARM computing resource pools is common, the current cloud platform lacks of cross-resource scheduling and management of the heterogeneous server resource pools, so that computing resource waste and computing service deployment density reduction are necessarily caused, and the technical trend and application requirements of computing resource on-demand scheduling and accurate matching in computing network time cannot be met. Therefore, the present embodiment provides a server resource scheduling management method to solve the above technical problems.
Referring to fig. 1, fig. 1 is a flowchart of a method for scheduling and managing server resources according to an embodiment of the present application.
As shown in fig. 1, the main steps of the method include steps S1 to S2.
Step S1, distributing the virtual machine to be deployed corresponding to the service to a corresponding target server resource pool for deployment according to the instruction set type and the hardware type of the service requirement.
And S2, if the hardware resources of the target server resource pool cannot meet the resource requirements of the virtual machines to be deployed, migrating the virtual machines in the target server resource pool to the heterogeneous server resource pool for deployment and/or restarting the virtual machines in the target server resource pool so as to release the hardware resources.
It should be noted that the server resource pools of different instruction sets are heterogeneous server resource pools. In this embodiment, a server resource scheduling management method between a heterogeneous X86 server resource pool and an ARM server resource pool is specifically described, and in this embodiment, if the target server resource pool is the X86 server resource pool, the heterogeneous server resource pool is the ARM server resource pool; if the target server resource pool is an ARM server resource pool, the heterogeneous server is an X86 server resource pool. An X86 cloud platform is deployed on the X86 server resource pool and used for scheduling and managing resources and virtual machines in the X86 server resource pool, and an ARM cloud platform is deployed on the ARM server resource pool and used for scheduling and managing resources and virtual machines in the ARM server resource pool. The server resource scheduling management method in the embodiment is implemented through a server resource scheduling management system as shown in fig. 2, the system is deployed in a server of a higher level of an X86 cloud platform and an ARM cloud platform and is deployed independently, the server resource scheduling management system can communicate and interact with controller nodes of the X86 cloud platform and the ARM cloud platform to communicate and manage the two cloud platforms, and further the cloud platforms schedule and manage resources and virtual machines of corresponding servers, so that the server resource scheduling management method in the embodiment is implemented.
Specifically, the allocating the virtual machine to be deployed corresponding to the service to the corresponding target server resource pool for deployment according to the instruction set type and the hardware type of the service requirement includes: if the service requires a specific type instruction set, distributing the virtual machine to be deployed to a target server resource pool of the specific type instruction set for deployment; if the service requires the specific type of hardware, the virtual machine to be deployed is distributed to a target server resource pool containing the specific type of hardware for deployment; if the service requires a specific type instruction set and specific type hardware, judging whether a target server resource pool of the specific type instruction set comprises the specific type hardware, if so, distributing the virtual machine to be deployed to the target server resource pool for deployment, otherwise, marking that the virtual machine to be deployed cannot be deployed, and sending alarm information of the lack of the specific type hardware in the target server resource pool; and if the service does not have a specific type instruction set and specific type hardware requirements, distributing the virtual machine to be deployed to a target server resource pool corresponding to the mirror image type of the virtual machine to be deployed.
The method comprises the steps of identifying and sensing issued services through an allocation module in a server resource scheduling management system, collecting whether the services require a specific type of instruction set and a specific type of hardware, and simultaneously collecting asset information of all PCIE computing cards in a server resource pool managed by the X86 cloud platform and the ARM cloud platform. Wherein the instruction set types include an X86 type instruction set and an ARM type instruction set, and the specific hardware includes a specific AI accelerator card, a GPU accelerator card, and the like.
For the situation of a specific type of instruction set of the service requirement, when the type of the instruction set of the service requirement is an X86 instruction set, configuring the virtual to be deployed corresponding to the service to be deployed only in an X86 resource pool, wherein the X86 server resource pool is a target server resource pool; when the instruction set type of the service requirement is ARM instruction set, the virtual machine to be deployed is configured to be deployed only in an ARM resource pool, and the ARM server resource pool is a target server resource pool.
Taking specific type hardware of service requirements as an AI accelerator card for example, when the total number of AI accelerators in an X86 resource pool is 0 and the total number of AI accelerators in an ARM resource pool is not 0, configuring a virtual machine to be deployed only in an ARM resource pool, wherein the ARM server resource pool is a target server resource pool; when the total number of the AI accelerator cards of the X86 resource pool is not 0 and the total number of the AI accelerator cards of the ARM resource pool is 0, configuring the virtual machine to be deployed only in the X86 resource pool, wherein the X86 server resource pool is a target server resource pool; and when the virtual machine to be deployed is 0, marking that the virtual machine to be deployed cannot be deployed, and giving alarm information on the system to prompt a manager to purchase a new hardware asset.
For the case that the service simultaneously requires a specific type instruction set and specific type hardware, taking the case that the service simultaneously requires an X86 instruction set and an AI accelerator card as an example, whether the AI accelerator card is included in an X86 server resource pool needs to be judged, if yes, the virtual machine to be deployed can be distributed to the X86 server resource pool for deployment, otherwise, the virtual machine to be deployed is marked to be incapable of deployment, and alarm information that the AI accelerator card is absent in the X86 server resource pool is sent out so as to remind an operator to add the AI accelerator card in the X86 server.
And for the situation when the service has no specific type instruction set and specific type hardware requirement, automatically judging to be deployed in the X86 or ARM resource pool according to the type of the virtual machine image to be deployed. When the service is issued, setting a mirror image file of the virtual machine to be deployed, and when the service does not have a specific type instruction set and a specific type hardware requirement, automatically distributing the virtual machine to be deployed to a corresponding resource pool for deployment according to the mirror image type of the virtual machine to be deployed. For example, if the type of the virtual machine image to be deployed is aaa-X86.Img, which is the X86 image, the X86 server resource pool is a target server resource pool; and if the type of the virtual machine image to be deployed is aaa-arm.img, which is the ARM image, the ARM server resource pool is a target server resource pool. Therefore, optimal matching of the service and the server resource is realized, and the deployment efficiency of the virtual machine in the heterogeneous cloud platform resource pool is improved.
Specifically, if the hardware resources of the target server resource pool cannot meet the resource requirements of the virtual machine to be deployed, migrating the virtual machine in the target server resource pool to the heterogeneous server resource pool for deployment and/or restarting the virtual machine in the target server resource pool to release the hardware resources, including: if the conventional hardware resources of the target server resource pool cannot meet the conventional hardware resource requirements of the virtual machine to be deployed, migrating the pure operation virtual machine in the target server resource pool to a heterogeneous server resource pool heterogeneous with the target server resource pool so as to release the conventional hardware resources; if the specific hardware resources in the target server resource pool are insufficient, restarting the virtual machine with the free specific hardware resources in the target server resource pool to release the specific hardware resources.
It should be understood that the hardware resources in the server resource pool include two types, one is a conventional hardware resource, which is a hardware resource for ensuring the normal operation of the server, and includes a CPU resource, a memory resource, a disk resource, and the like; another is a hardware resource with special power functions, such as AI accelerator card resources for big data analysis and GPU accelerator card resources for image processing, etc. The purely operational virtual machine in this embodiment is a virtual machine that has no specific type of hardware requirements.
Preferably, if the conventional hardware resources of the target server resource pool cannot meet the conventional hardware resource requirements of the virtual machine to be deployed, migrating the pure operation virtual machine in the target server resource pool to a heterogeneous server resource pool heterogeneous with the target server resource pool to release the conventional hardware resources, including: according to the demand of the virtual machine to be deployed for each conventional hardware resource, determining a first conventional hardware resource which cannot meet the demand of the virtual machine to be deployed in a target server resource pool; determining the occupation amount of each pure operation virtual machine in a target server resource pool to each conventional hardware resource; multiplying the occupation amount of each pure operation virtual machine to each conventional hardware resource by a weight coefficient corresponding to each conventional hardware resource, and adding the results to obtain a migration coefficient of each pure operation virtual machine, wherein the weight coefficient of the first conventional hardware resource is the largest; determining whether a pure operation virtual machine with the occupation amount of the first conventional hardware resources being larger than or equal to the first conventional demand amount of the virtual machine to be deployed for the first conventional hardware resources exists; if yes, the pure operation virtual machine with the first conventional hardware resource occupation amount being larger than or equal to the first conventional demand amount and the minimum migration coefficient is used as a migration virtual machine and migrated to the heterogeneous server resource pool; otherwise, the plurality of pure operation virtual machines with the minimum sum of migration coefficients are taken as migration virtual machines and migrated to the heterogeneous server resource pool, wherein the sum of the first conventional hardware resource occupation amount is larger than or equal to the first conventional demand amount.
The allocation module of the server resource scheduling management system issues the X86 or ARM cloud platform to perform virtual machine deployment after the virtual machine to be deployed is allocated to a proper resource pool for deployment. The present embodiment is described taking the target server resource pool as the X86 server resource pool as an example, as shown in fig. 3. When the virtual machine to be deployed of the X86 cloud platform cannot be deployed, detecting a creation failure reason of the virtual machine to be deployed, which is reported by the X86 cloud platform, through a resource management module in a server resource scheduling management system. The main reasons for the failure of the creation of the virtual machine caused by the shortage of the conventional hardware resources include three reasons of insufficient CPU resources, insufficient memory resources and insufficient disk resources. And inquiring configuration XML files of all virtual machines currently running in the X86 resource pool from the X86 cloud platform through the resource management module, wherein the virtual machines without special hardware PCIE card configuration are judged to be purely operational virtual machines. And recording the occupation amount of each pure operation virtual machine on each conventional hardware resource, wherein the occupied CPU core number is recorded as m, the occupied memory capacity is recorded as l, the occupied disk capacity is recorded as n, and the weight coefficient of the CPU resource is set as k1, the weight coefficient of the memory resource is set as k2 and the weight coefficient of the disk resource is set as k3.
When the virtual machine fails to be created due to insufficient CPU resources, the weight coefficient k1 of the CPU resources is set to be maximum, and k1=0.6, k2=0.3, and k3=0.1 are set in this embodiment. And calculating migration coefficients of each pure operation virtual machine according to a formula w=k1+k2+k3+k3, wherein W is the migration coefficient of the pure operation virtual machine. Comparing the number of CPU cores occupied by each pure operation virtual machine in the X86 server resource pool with the number P of CPU cores required by the virtual machine to be deployed. If the CPU core number P occupied by the pure operation virtual machine is larger than or equal to the CPU core number required by the virtual machine to be deployed, selecting the pure operation virtual machine with the CPU core number larger than or equal to P and the minimum migration coefficient W as a migration virtual machine, and migrating the migration virtual machine into a heterogeneous resource pool to release the CPU resources occupied by the migration virtual machine, so that the virtual machine to be deployed is successfully deployed. When the number of CPU cores occupied by all the pure operation virtual machines in the X86 server resource pool is smaller than P, a plurality of pure operation virtual machines are required to be migrated, the sum of the number of the occupied CPU cores is larger than or equal to P, and a plurality of pure operation virtual machines with the minimum sum of migration coefficients W are taken as migration virtual machines, and the migration virtual machines are migrated to a heterogeneous resource pool to release CPU resources.
When the virtual machine fails to be created due to insufficient memory resources, the weight coefficient k2 of the memory resources is set to be the largest, k1=0.3, k2=0.6 and k3=0.1. The migration coefficients of each purely operational virtual machine are also calculated according to the formula w=k1×m+k2×l+k3×n. And comparing the memory capacity occupied by each pure operation virtual machine in the X86 server resource pool with the memory capacity M required by the virtual machine to be deployed. If the memory capacity occupied by the pure operation virtual machine is larger than or equal to the memory capacity M required by the virtual machine to be deployed, selecting the pure operation virtual machine with the memory capacity larger than or equal to M and the minimum migration coefficient W as a migration virtual machine, and migrating the migration virtual machine into a heterogeneous resource pool to release memory resources. When the memory capacity occupied by all the pure operation virtual machines in the X86 server resource pool is smaller than M, the method indicates that the migration needs to be carried out on a plurality of pure operation virtual machines, the sum of the occupied memory capacities is larger than or equal to P, and a plurality of pure operation virtual machines with the minimum sum of migration coefficients W are used as migration virtual machines, and the migration virtual machines are migrated into a heterogeneous resource pool so as to release memory resources.
When the virtual machine fails to be created due to insufficient disk resources, the set weight coefficient k3 of the disk resources is set to be the largest, k1=0.1, k2=0.3, and k3=0.6. The migration coefficients of each purely operational virtual machine are also calculated according to the formula w=k1×m+k2×l+k3×n. And comparing the disk capacity occupied by each pure operation virtual machine in the target server resource pool with the disk capacity D required by the virtual machine to be deployed. If the disk capacity occupied by the pure operation virtual machine is larger than or equal to the disk capacity D required by the virtual machine to be deployed, selecting the pure operation virtual machine with the disk capacity occupied larger than or equal to D and the minimum migration coefficient W as a migration virtual machine, and migrating the migration virtual machine into a heterogeneous resource pool to release disk resources. When the disk capacity occupied by all the pure operation virtual machines in the target server resource pool is smaller than D, the fact that the plurality of pure operation virtual machines need to be migrated is indicated, the sum of the occupied disk capacities is larger than or equal to that of the plurality of pure operation virtual machines, and the sum of migration coefficients W is minimum, the plurality of pure operation virtual machines are used as migration virtual machines, and the migration virtual machines are migrated to the heterogeneous resource pool so as to release disk resources.
It should be noted that, if the CPU resources, the memory resources, or the disk resources in the heterogeneous resource pool are insufficient to deploy the migration virtual machine, the server resource scheduling management system feeds back that the heterogeneous virtual machine cannot receive the information of the migration virtual machine, and stops migrating the virtual machine.
Further, after determining to migrate the virtual machine, migrating the migrating virtual machine to the heterogeneous server resource pool, including: inquiring the instruction set type, format and disk size of the migration virtual machine; determining whether the instruction set is a specific type instruction set corresponding to the target server resource pool according to the instruction set type of the migration virtual machine; if yes, canceling migration of the migration virtual machine, and returning migration failure information; otherwise, creating an empty mirror image file in the heterogeneous server resource pool, and configuring a new virtual machine according to the format and the disk size of the virtual machine to be migrated; and translating the format of the instruction set of the virtual machine to be migrated into the format of the instruction set of the heterogeneous server resource pool, and writing the translated instruction set into the empty mirror image file.
By way of example, the instruction set type, the instruction set format and the disk size of the migration virtual machine image file are queried through the qemu info command of the virtual machine image management program, a new empty image file is created through the qemu-img create command of the virtual machine image management program according to the format and the disk size, a virtual machine with the simplest configuration is started through the qemu command line in the easily purchased server resource pool to be migrated, the virtual machine is fixedly configured into 1 core and 512M memory, and then assembly instructions in the qemu executing process are captured, translated and replaced through a linux tool. In the process, if the migration virtual machine is found to have a specific instruction set corresponding to the target server resource pool, the mirror image can only be deployed in the target server resource pool, the mirror image conversion can not be performed, the migration virtual machine can not be migrated to the heterogeneous resource pool, and the program is exited. If the migration virtual machine has no specific instruction set requirement, writing a new instruction set into an empty mirror image file through a linux tool, and completing conversion of the virtual machine mirror image.
In some embodiments, the method further comprises: determining the occupancy rate of each specific hardware resource in each current resource pool according to the total number of PCIE cards in each specific hardware of each server resource pool, the set maximum power consumption of the PCIE cards and the actual total power consumption of the current PCIE cards; and if the occupancy rate of the specific hardware resources is larger than or equal to a preset occupancy rate threshold value, sending out a warning of insufficient specific hardware resources.
The resource management module in this embodiment further monitors the resource occupancy rate of each specific hardware in each resource pool. And inquiring hardware asset information stored in the X86 and ARM cloud platforms through a resource management module, determining the maximum power consumption T of PCIE cards in each specific hardware, the total number N of PCIE cards in each hardware and the current actual total power consumption M of each PCIE card, and calculating the resource occupancy rate E of each PCIE card according to a formula E=M/(T.times.N). It should be understood that the power of the interface of the specific hardware may represent the usage of specific hardware resources such as the AI accelerator card and the GPU accelerator card. When the occupancy rate of the resources is greater than or equal to a preset occupancy rate threshold value Y, a corresponding alarm of insufficient specific hardware resources is generated to prompt a resource pool operation and maintenance personnel, so that the hardware resources are insufficient, and the purchasing and deployment of the hardware resources can be properly increased; otherwise, purchasing and deploying the hardware resource are not needed. Therefore, the monitoring of specific hardware resources is realized, and the user is supported to configure the specific hardware model and the alarm threshold of the monitoring.
Preferably, if the specific hardware resources in the target server resource pool are insufficient, restarting the virtual machine with the free specific hardware resources in the target server resource pool to release the specific hardware resources, including: determining the idle quantity of the first specific hardware resources in each virtual machine according to the occupation quantity of the first specific hardware resources with insufficient resources of each virtual machine in the target resource pool, the real-time power consumption of the PCIE card in the first specific hardware and the set maximum power consumption of the PCIE card; determining whether there is a virtual machine having an amount of idleness for the first specific hardware resource greater than or equal to a first specific demand for the first specific hardware resource by the virtual machine to be deployed; if yes, taking the virtual machine with the largest first specific hardware resource idle amount and larger than or equal to the first specific demand as a virtual machine to be restarted, configuring the occupation amount of the first specific hardware resource of the virtual machine to be restarted as the occupation amount of the first specific hardware resource before restarting minus the first specific demand, and restarting the virtual machine to be restarted to release the first specific hardware resource; otherwise, taking the plurality of virtual machines with the sum of the idle amounts of the first specific hardware resources being greater than or equal to the first specific demand as virtual machines to be restarted, configuring the total occupied amount of the first specific hardware resources of the plurality of virtual machines to be restarted as the total occupied amount of the first specific hardware resources before the virtual machines are restarted minus the first specific demand, and restarting the virtual machines to be restarted to release the first specific hardware resources.
The specific hardware resources are typically offloaded by a resource management module in the server resource scheduling management system. In this embodiment, the case that the target server resource pool is an X86 server resource pool and the AI acceleration card in the X86 server resource pool is insufficient when the virtual machine to be deployed is described as an example.
When AI acceleration card resources in the X86 server resource pool are insufficient, determining the required number R of AI acceleration card resources of the virtual machines to be deployed, inquiring all virtual machines currently configured with the AI acceleration card resources in the resource pool from a cloud platform corresponding to the X86 server resource pool by a resource management module, inquiring the AI acceleration card resource number A configured by each virtual machine, collecting real-time power consumption B of PCIE cards corresponding to AI technology cards configured by each virtual machine, and calculating the number F of idle AI acceleration card resources in each virtual machine according to a formula F=A-B/T. And the number of idle AI acceleration card resources is the largest, and the virtual machines meeting the requirement of (A-B/T) not less than R are the virtual machines to be restarted. At the moment, a command for modifying the AI accelerator card resource configuration of the virtual machine to be restarted is initiated to the cloud platform, the number of AI accelerator card resources of the virtual machine to be restarted is configured to be A-R, and a command for restarting the virtual machine to be restarted is initiated to the cloud platform by the resource management module after the configuration is completed. After the restart of the virtual machine is completed, the AI acceleration card resource can be unloaded, and the AI acceleration card resource is released. So that a new virtual machine can be created at this resource pool. When the number of the AI accelerator card resources in all the virtual machines in the target server resource pool is smaller than R, the AI accelerator card resources in the virtual machines need to be reconfigured and restarted, and the total number of the AI accelerator card resources of the restarting virtual machines is configured to be the total occupation amount of the first specific hardware resources before the restarting minus the first specific demand amount A-R. The method realizes the automatic detection and unloading of specific hardware resources, and can effectively improve the concentration of the deployed virtual machines in the server resource pool.
When the target server resource pool is an ARM server resource pool and the X86 server resource pool is a heterogeneous server resource pool, the server resource scheduling management method is similar in flow, and the specific flow is shown in fig. 4 and will not be described again.
The server resource scheduling management method and system provided by the embodiment of the application have the beneficial effects that: the application increases the service demand sensing capability for the cloud platform, can automatically detect the special instruction set demand and the special computing power hardware resource demand of the service, and then allocates a resource pool for the service according to the detection result, thereby ensuring the allocation rationality when the service allocates the resource pool. When the resources in the resource pool of the target server can not meet the service requirements, the pure operation virtual machine is migrated to the heterogeneous resource pool for deployment, so that unified management of resources in the heterogeneous X86 resource pool and ARM resource pool is realized, the utilization rate of computing resources is improved, and load balancing is realized. And converting the service virtual machine mirror image of the heterogeneous resource cloud platform, so that deployment is performed on the heterogeneous resource cloud platform. When the virtual machine is deployed, the occupation condition of the hardware resources of the virtual machine deployed in the server resource pool is detected, and the virtual machine with idle hardware resources is reconfigured and restarted to release the hardware resources, so that the virtual machine to be deployed is successfully deployed, the deployment density of the virtual machine in the resource pool is effectively improved, and the problem that specific hardware resources are blindly purchased to cause specific hardware computing power resource waste is avoided.
Referring to fig. 5, fig. 5 is a schematic block diagram of a server resource scheduling management system according to an embodiment of the present application.
As shown in fig. 5, the system includes:
the allocation module is used for allocating the virtual machine to be deployed corresponding to the service to a corresponding target server resource pool for deployment according to the instruction set type and the hardware type of the service requirement;
and the resource management module is used for migrating the virtual machines in the target server resource pool to the heterogeneous server resource pool for deployment and/or restarting the virtual machines in the target server resource pool to release the hardware resources if the hardware resources of the target server resource pool cannot meet the resource requirements of the virtual machines to be deployed.
The server resource scheduling management system is arranged in a cloud platform corresponding to the target server resource pool and a superior server of the cloud platform corresponding to the heterogeneous resource pool.
Wherein the resource management module is further configured to:
if the conventional hardware resources of the target server resource pool cannot meet the conventional hardware resource requirements of the virtual machine to be deployed, migrating the pure operation virtual machine in the target server resource pool to a heterogeneous server resource pool heterogeneous with the target server resource pool so as to release the conventional hardware resources;
Restarting the virtual machine with idle specific hardware resources in the target server resource pool to release the specific hardware resources if the specific hardware resources in the target server resource pool are insufficient;
the conventional hardware resources comprise CPU resources, memory resources and disk resources, and the specific hardware resources comprise AI acceleration card resources and GPU acceleration card resources;
the pure operation virtual machine is a virtual machine without specific type of hardware requirements.
Wherein the resource management module is further configured to:
according to the demand of the virtual machine to be deployed for each conventional hardware resource, determining a first conventional hardware resource which cannot meet the demand of the virtual machine to be deployed in a target server resource pool;
determining the occupation amount of each pure operation virtual machine in a target server resource pool to each conventional hardware resource;
multiplying the occupation amount of each pure operation virtual machine to each conventional hardware resource by a weight coefficient corresponding to each conventional hardware resource, and adding the results to obtain a migration coefficient of each pure operation virtual machine, wherein the weight coefficient of the first conventional hardware resource is the largest;
determining whether a pure operation virtual machine with the occupation amount of the first conventional hardware resources being larger than or equal to the first conventional demand amount of the virtual machine to be deployed for the first conventional hardware resources exists;
If yes, the pure operation virtual machine with the first conventional hardware resource occupation amount being larger than or equal to the first conventional demand amount and the minimum migration coefficient is used as a migration virtual machine and migrated to the heterogeneous server resource pool;
otherwise, the plurality of pure operation virtual machines with the minimum sum of migration coefficients are taken as migration virtual machines and migrated to the heterogeneous server resource pool, wherein the sum of the first conventional hardware resource occupation amount is larger than or equal to the first conventional demand amount.
Wherein the resource management module is further configured to:
inquiring the instruction set type, format and disk size of the migration virtual machine;
determining whether the instruction set is a specific type instruction set corresponding to the target server resource pool according to the instruction set type of the migration virtual machine;
if yes, canceling migration of the migration virtual machine, and returning migration failure information;
otherwise, creating an empty mirror image file in the heterogeneous server resource pool, and configuring a new virtual machine according to the format and the disk size of the virtual machine to be migrated;
and translating the format of the instruction set of the virtual machine to be migrated into the format of the instruction set of the heterogeneous server resource pool, and writing the translated instruction set into the empty mirror image file.
Wherein the resource management module is further configured to:
determining the occupancy rate of each specific hardware resource in each current resource pool according to the total number of PCIE cards in each specific hardware of each server resource pool, the set maximum power consumption of the PCIE cards and the actual total power consumption of the current PCIE cards;
and if the occupancy rate of the specific hardware resources is larger than or equal to a preset occupancy rate threshold value, sending out a warning of insufficient specific hardware resources.
Wherein the resource management module is further configured to:
determining the idle quantity of the first specific hardware resources in each virtual machine according to the occupation quantity of the first specific hardware resources with insufficient resources of each virtual machine in the target resource pool, the real-time power consumption of the PCIE card in the first specific hardware and the set maximum power consumption of the PCIE card;
determining whether there is a virtual machine having an amount of idleness for the first specific hardware resource greater than or equal to a first specific demand for the first specific hardware resource by the virtual machine to be deployed;
if yes, taking the virtual machine with the largest first specific hardware resource idle amount and larger than or equal to the first specific demand as a virtual machine to be restarted, configuring the occupation amount of the first specific hardware resource of the virtual machine to be restarted as the occupation amount of the first specific hardware resource before restarting minus the first specific demand, and restarting the virtual machine to be restarted to release the first specific hardware resource;
Otherwise, taking the plurality of virtual machines with the sum of the idle amounts of the first specific hardware resources being greater than or equal to the first specific demand as virtual machines to be restarted, configuring the total occupied amount of the first specific hardware resources of the plurality of virtual machines to be restarted as the total occupied amount of the first specific hardware resources before the virtual machines are restarted minus the first specific demand, and restarting the virtual machines to be restarted to release the first specific hardware resources.
Wherein the allocation module is further configured to:
if the service requires a specific type instruction set, distributing the virtual machine to be deployed to a target server resource pool of the specific type instruction set for deployment;
if the service requires the specific type of hardware, the virtual machine to be deployed is distributed to a target server resource pool containing the specific type of hardware for deployment;
if the service requires a specific type instruction set and specific type hardware, judging whether a target server resource pool of the specific type instruction set comprises the specific type hardware, if so, distributing the virtual machine to be deployed to the target server resource pool for deployment, otherwise, marking that the virtual machine to be deployed cannot be deployed, and sending alarm information of the lack of the specific type hardware in the target server resource pool;
And if the service does not have a specific type instruction set and specific type hardware requirements, distributing the virtual machine to be deployed to a target server resource pool corresponding to the mirror image type of the virtual machine to be deployed.
The server resource pools with different instruction sets are heterogeneous server resource pools.
It should be noted that, for convenience and brevity of description, specific working procedures of the above-described apparatus and each module and unit may refer to corresponding procedures in the foregoing embodiments, and are not repeated herein.
It should be noted that, in this document, the terms "comprises," "comprising," or any other variation thereof, are intended to cover a non-exclusive inclusion, such that a process, method, article, or system that comprises a list of elements does not include only those elements but may include other elements not expressly listed or inherent to such process, method, article, or system. Without further limitation, an element defined by the phrase "comprising one … …" does not exclude the presence of other like elements in a process, method, article, or system that comprises the element.
The foregoing embodiment numbers of the present application are merely for the purpose of description, and do not represent the advantages or disadvantages of the embodiments. While the application has been described with reference to certain preferred embodiments, it will be understood by those skilled in the art that various changes and substitutions of equivalents may be made and equivalents will be apparent to those skilled in the art without departing from the scope of the application. Therefore, the protection scope of the application is subject to the protection scope of the claims.

Claims (10)

1. A server resource scheduling management method, comprising:
distributing the virtual machines to be deployed corresponding to the service to a corresponding target server resource pool for deployment according to the instruction set type and the hardware type of the service requirement;
if the hardware resources of the target server resource pool cannot meet the resource requirements of the virtual machines to be deployed, migrating the virtual machines in the target server resource pool to the heterogeneous server resource pool for deployment and/or restarting the virtual machines in the target server resource pool so as to release the hardware resources.
2. The method of claim 1, wherein if the hardware resources of the target server resource pool cannot meet the resource requirements of the virtual machine to be deployed, migrating the virtual machine in the target server resource pool to the heterogeneous server resource pool for deployment and/or restarting the virtual machine in the target server resource pool to release the hardware resources, comprising:
If the conventional hardware resources of the target server resource pool cannot meet the conventional hardware resource requirements of the virtual machine to be deployed, migrating the pure operation virtual machine in the target server resource pool to a heterogeneous server resource pool heterogeneous with the target server resource pool so as to release the conventional hardware resources;
restarting the virtual machine with idle specific hardware resources in the target server resource pool to release the specific hardware resources if the specific hardware resources in the target server resource pool are insufficient;
the conventional hardware resources comprise CPU resources, memory resources and disk resources, and the specific hardware resources comprise AI acceleration card resources and GPU acceleration card resources;
the pure operation virtual machine is a virtual machine without specific type of hardware requirements.
3. The method according to claim 2, wherein if the regular hardware resources of the target server resource pool cannot meet the regular hardware resource requirements of the virtual machine to be deployed, migrating the pure operation virtual machine in the target server resource pool to a heterogeneous server resource pool heterogeneous to the target server resource pool to release the regular hardware resources, comprising:
According to the demand of the virtual machine to be deployed for each conventional hardware resource, determining a first conventional hardware resource which cannot meet the demand of the virtual machine to be deployed in a target server resource pool;
determining the occupation amount of each pure operation virtual machine in a target server resource pool to each conventional hardware resource;
multiplying the occupation amount of each pure operation virtual machine to each conventional hardware resource by a weight coefficient corresponding to each conventional hardware resource, and adding the results to obtain a migration coefficient of each pure operation virtual machine, wherein the weight coefficient of the first conventional hardware resource is the largest;
determining whether a pure operation virtual machine with the occupation amount of the first conventional hardware resources being larger than or equal to the first conventional demand amount of the virtual machine to be deployed for the first conventional hardware resources exists;
if yes, the pure operation virtual machine with the first conventional hardware resource occupation amount being larger than or equal to the first conventional demand amount and the minimum migration coefficient is used as a migration virtual machine and migrated to the heterogeneous server resource pool;
otherwise, the plurality of pure operation virtual machines with the minimum sum of migration coefficients are taken as migration virtual machines and migrated to the heterogeneous server resource pool, wherein the sum of the first conventional hardware resource occupation amount is larger than or equal to the first conventional demand amount.
4. The server resource scheduling management method of claim 3, wherein migrating the migrated virtual machine into the heterogeneous server resource pool comprises:
inquiring the instruction set type, format and disk size of the migration virtual machine;
determining whether the instruction set is a specific type instruction set corresponding to the target server resource pool according to the instruction set type of the migration virtual machine;
if yes, canceling migration of the migration virtual machine, and returning migration failure information;
otherwise, creating an empty mirror image file in the heterogeneous server resource pool, and configuring a new virtual machine according to the format and the disk size of the virtual machine to be migrated;
and translating the format of the instruction set of the virtual machine to be migrated into the format of the instruction set of the heterogeneous server resource pool, and writing the translated instruction set into the empty mirror image file.
5. The server resource scheduling management method according to claim 2, characterized in that the method further comprises:
determining the occupancy rate of each specific hardware resource in each current resource pool according to the total number of PCIE cards in each specific hardware of each server resource pool, the set maximum power consumption of the PCIE cards and the actual total power consumption of the current PCIE cards;
And if the occupancy rate of the specific hardware resources is larger than or equal to a preset occupancy rate threshold value, sending out a warning of insufficient specific hardware resources.
6. The method for scheduling and managing server resources according to claim 5, wherein restarting the virtual machine with the idle specific hardware resources in the target server resource pool to release the specific hardware resources if the specific hardware resources in the target server resource pool are insufficient, comprises:
determining the idle quantity of the first specific hardware resources in each virtual machine according to the occupation quantity of the first specific hardware resources with insufficient resources of each virtual machine in the target resource pool, the real-time power consumption of the PCIE card in the first specific hardware and the set maximum power consumption of the PCIE card;
determining whether there is a virtual machine having an amount of idleness for the first specific hardware resource greater than or equal to a first specific demand for the first specific hardware resource by the virtual machine to be deployed;
if yes, taking the virtual machine with the largest first specific hardware resource idle amount and larger than or equal to the first specific demand as a virtual machine to be restarted, configuring the occupation amount of the first specific hardware resource of the virtual machine to be restarted as the occupation amount of the first specific hardware resource before restarting minus the first specific demand, and restarting the virtual machine to be restarted to release the first specific hardware resource;
Otherwise, taking the plurality of virtual machines with the sum of the idle amounts of the first specific hardware resources being greater than or equal to the first specific demand as virtual machines to be restarted, configuring the total occupied amount of the first specific hardware resources of the plurality of virtual machines to be restarted as the total occupied amount of the first specific hardware resources before the virtual machines are restarted minus the first specific demand, and restarting the virtual machines to be restarted to release the first specific hardware resources.
7. The method for scheduling and managing server resources according to claim 1, wherein the allocating the virtual machine to be deployed corresponding to the service to the corresponding target server resource pool for deployment according to the instruction set type and the hardware type of the service requirement includes:
if the service requires a specific type instruction set, distributing the virtual machine to be deployed to a target server resource pool of the specific type instruction set for deployment;
if the service requires the specific type of hardware, the virtual machine to be deployed is distributed to a target server resource pool containing the specific type of hardware for deployment;
if the service requires a specific type instruction set and specific type hardware, judging whether a target server resource pool of the specific type instruction set comprises the specific type hardware, if so, distributing the virtual machine to be deployed to the target server resource pool for deployment, otherwise, marking that the virtual machine to be deployed cannot be deployed, and sending alarm information of the lack of the specific type hardware in the target server resource pool;
And if the service does not have a specific type instruction set and specific type hardware requirements, distributing the virtual machine to be deployed to a target server resource pool corresponding to the mirror image type of the virtual machine to be deployed.
8. The method for scheduling and managing server resources according to claim 1, wherein,
the server resource pools of different instruction sets are heterogeneous server resource pools.
9. A server resource scheduling management system, characterized in that it is configured to implement a server resource scheduling management method according to any one of claims 1 to 8, the system comprising:
the allocation module is used for allocating the virtual machine to be deployed corresponding to the service to a corresponding target server resource pool for deployment according to the instruction set type and the hardware type of the service requirement;
and the resource management module is used for migrating the virtual machines in the target server resource pool to the heterogeneous server resource pool for deployment and/or restarting the virtual machines in the target server resource pool to release the hardware resources if the hardware resources of the target server resource pool cannot meet the resource requirements of the virtual machines to be deployed.
10. The server resource scheduling management system according to claim 9, wherein the server resource scheduling management system is disposed in a server of a higher level of a cloud platform corresponding to the target server resource pool and a cloud platform corresponding to the heterogeneous resource pool.
CN202310910792.5A 2023-07-21 2023-07-21 Method and system for scheduling and managing server resources Pending CN116974762A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202310910792.5A CN116974762A (en) 2023-07-21 2023-07-21 Method and system for scheduling and managing server resources

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202310910792.5A CN116974762A (en) 2023-07-21 2023-07-21 Method and system for scheduling and managing server resources

Publications (1)

Publication Number Publication Date
CN116974762A true CN116974762A (en) 2023-10-31

Family

ID=88470699

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202310910792.5A Pending CN116974762A (en) 2023-07-21 2023-07-21 Method and system for scheduling and managing server resources

Country Status (1)

Country Link
CN (1) CN116974762A (en)

Similar Documents

Publication Publication Date Title
US9319281B2 (en) Resource management method, resource management device, and program product
US9727355B2 (en) Virtual Hadoop manager
JP5417287B2 (en) Computer system and computer system control method
US20170017511A1 (en) Method for memory management in virtual machines, and corresponding system and computer program product
CN111506423B (en) Resource scheduling method and device, electronic equipment and storage medium
CN105378669A (en) Virtual machine resource management system and method thereof
CN111190691A (en) Automatic migration method, system, device and storage medium suitable for virtual machine
CN112988361B (en) Cluster task allocation method and device and computer readable medium
CN117480494A (en) Coordinated container scheduling for improved resource allocation in virtual computing environments
CN111638953B (en) Method, device and storage medium for realizing GPU virtualization
CN114968601B (en) Scheduling method and scheduling system for AI training jobs with resources reserved in proportion
CN111857951A (en) Containerized deployment platform and deployment method
CN112256430A (en) Container deployment method, device, equipment and storage medium
CN111796933A (en) Resource scheduling method, device, storage medium and electronic equipment
CN109347716B (en) Instantiation method and device of consumer VNF
CN114546587A (en) Capacity expansion and reduction method of online image recognition service and related device
CN112003931B (en) Method and system for deploying scheduling controller and related components
CN111768106B (en) Elevator resource allocation method, system, electronic equipment and storage medium
CN117435324A (en) Task scheduling method based on containerization
CN116974762A (en) Method and system for scheduling and managing server resources
CN109062580B (en) Virtual environment deployment method and deployment device
CN107562510B (en) Management method and management equipment for application instances
CN115495239A (en) Test resource multiplexing method and device, storage medium and computer equipment
CN112612579A (en) Virtual machine deployment method, storage medium, and computer device
CA2313273A1 (en) Device and method for allocating jobs in a network

Legal Events

Date Code Title Description
PB01 Publication
PB01 Publication
SE01 Entry into force of request for substantive examination
SE01 Entry into force of request for substantive examination