CN104899094A - Virtual machine load balance processing method - Google Patents

Virtual machine load balance processing method Download PDF

Info

Publication number
CN104899094A
CN104899094A CN201410077897.8A CN201410077897A CN104899094A CN 104899094 A CN104899094 A CN 104899094A CN 201410077897 A CN201410077897 A CN 201410077897A CN 104899094 A CN104899094 A CN 104899094A
Authority
CN
China
Prior art keywords
virtual machine
server
management
administration module
domain
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
CN201410077897.8A
Other languages
Chinese (zh)
Inventor
金飞
葛菲
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.)
Anhui Wanxin Wisedu Technology Co Ltd
Original Assignee
Anhui Wanxin Wisedu 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 Anhui Wanxin Wisedu Technology Co Ltd filed Critical Anhui Wanxin Wisedu Technology Co Ltd
Priority to CN201410077897.8A priority Critical patent/CN104899094A/en
Publication of CN104899094A publication Critical patent/CN104899094A/en
Pending legal-status Critical Current

Links

Landscapes

  • Computer And Data Communications (AREA)

Abstract

The invention relates to the technical field of a data center, in particular to a virtual machine load balance processing method. The virtual machine load balance processing method is applied to a main management module of a virtual machine server cluster domain; virtual machine servers of different manufacturers are mounted in the virtual machine server cluster domain; the virtual machine servers of each manufacturer belong to the management of management modules of the manufacturer per se; a management module in charge of the load balance management of the cluster domain is the main management module, the rest of management modules are slave management modules; virtual machine resource information and virtual machine load information in management regions of the management modules are collected by communication with the virtual machine servers managed by the main management module; the virtual machine servers inquire virtual machines (VM) which are in a useful state and have the minimal external movable connection number; and if only one VM which is in the useful state and has the minimal external movable connection number is inquired, the VM is used as a target VM, and the target VM directly issues to the corresponding virtual machine server.

Description

A kind of method of virtual machine load balance process
Technical field
The present invention relates to the technical field of data center, particularly relate to a kind of method of virtual machine load balance process.
Background technology
Load balancing (Load Balance), also referred to as load balancing, load (task) can be carried out balancing, share on multiple operating unit and perform, such as Web server, FTP server, enterprise crucial application server and other mission critical server etc., thus task of jointly finishing the work.Wherein, virtual machine (Virtual Machine) refer to by software simulation have complete hardware system function, the complete computer operated in a completely isolated environment.Virtual machine load-balancing technique ceaselessly can monitor the utility ratio of resource pool in virtual server mainframe cluster continuously, and can distribute resource needed for it intelligently according to service needed in virtual machine.By dynamic distribution EQUILIBRIUM CALCULATION FOR PROCESS resource, virtual machine load-balancing technique can integrated service device, thus the continuity of the business of maintenance and security, raising resource utilization, save IT cost and energy resource consumption.
Intel Virtualization Technology is the core key technology in cloud computing.One station server can be virtualized into multiple stage virtual machine VM (Virtual Machine), for user provides more efficient data, services by Intel Virtualization Technology.But, due to the difference of each physical host processing power and the uncertainty of each application program, result in the load imbalance of each physical host in virtual computation environmental, namely the resource utilization of some physical host is very high, cause the resource of this physical host depleted, and the resource utilization of some physical host is very low, even idle, do not effectively utilized.
In current virtualized mainstream vendor, VMware and Microsoft has released virtual machine load-balancing technique, and the virtual XEN that increases income also releases virtual load-balancing technique in preparation.Wherein, different manufacturers utilizes virtual machine management platform to carry out the instrument of load balancing, strategy, communication mode are normally different, such as, by the loading condition weighing whole cluster environment, the load balancing operating strategy of VMware judges whether current hosts transships, in order to reach balance, load balancing management moves out virtual machine by virtual machine migration tool from the main frame of overload.And the load balancing operating strategy of Microsoft can not remove the loading condition calculating whole cluster, resource utilization by means of only separate unit host decides to do online migration the need of to virtual machine after whether this main frame transships and transship, such as, if when the memory usage of current hosts exceedes " physical memory total amount deducts the reserved amount of ram of host ", load balancing service just thinks that this main frame transships.Therefore, in prior art, the mutual load of virtual machine server in cluster that the mixing cluster installed based on the virtual machine server of different manufacturers does not realize different manufacturers is shared, and therefore, the existing management of the virtual machine load balancing based on cluster haves much room for improvement.
Summary of the invention
The present invention is for avoiding the weak point that exists in above-mentioned prior art, a kind of method of virtual machine load balance process is provided, the virtual machine server solving different manufacturers can not cooperate and share in virtual machine server cluster of domain, to realize the balance dispatching to VM resource, make server resource can obtain the scheduling of maximal efficiency.
Technical scheme of the present invention is: a kind of method of virtual machine load balance process, be applied in the main management module of virtual machine server cluster of domain, described virtual machine server cluster of domain has installed the virtual machine server of different manufacturers, the virtual machine server of each producer belongs to the administration module management of this producer, the administration module being responsible for the management of cluster of domain load balancing is main management module, remaining administration module is from administration module, comprises the following steps:
A. the resources of virtual machine information in this administration module management domain and virtual machine load information is collected by communicating with the virtual machine server that this administration module manages, from the resources of virtual machine information in administration module management domain and virtual machine load information described in reception reports from administration module, add up the resources of virtual machine information in described virtual machine server cluster of domain and virtual machine load information;
B. virtual machine server inquiry is in upstate and is externally flexibly connected the minimum virtual machine VM of number; If only inquire a virtual machine VM to be in upstate and to be externally flexibly connected number minimum, then using this virtual machine VM as target virtual machine VM;
Be in upstate more than the virtual machine VM of and to be externally flexibly connected number minimum if c. inquired, then in this multiple stage virtual machine VM, choose the minimum virtual machine VM of real-time traffic as target virtual machine VM;
D. on connection establishment to target virtual machine VM, respective virtual machine server is directly handed down to by this target virtual machine VM, as produced the load balancing task of virtual machine server from administration module management domain, then send to and issue from administration module accordingly.
In described step c, during virtual machine server inquiry virtual machine VM, if when a certain virtual machine VM is in and carries out the transition state moved between different server, the virtual machine VM that this is in transition state by virtual machine server is set to down state; After migration completes, again this virtual machine VM is set to upstate, virtual machine server adds up the external flexible connection number of this virtual machine VM, makes this virtual machine VM again participate in load balance scheduling.
Compared with the prior art, beneficial effect of the present invention is embodied in:
In the method for virtual machine load balance process of the present invention, be responsible for the resources of virtual machine information of virtual machine server in this management domain and the collection of virtual machine load information from administration module and report, and be responsible for sharing issuing of task, the virtual machine server achieving different manufacturers by the cooperation between master and slave administration module cooperates load sharing in cluster of domain, improves resource utilization and the scheduling of resource ability of data center's virtual machine server cluster.By the scheduling of this combined strategy, external request can be evenly distributed on each VM, thus realize the balance dispatching to VM resource, make server resource can obtain the scheduling of maximal efficiency.] Real-Time Scheduling of the present invention, server resource expense is less, and efficiency is high.The inventive method, mutually compatible with existing standard, dispatching algorithm efficiency is high, and load balancing scheduling of resource under virtualized environment is evenly distributed.
Embodiment
Be applied in the main management module of virtual machine server cluster of domain, described virtual machine server cluster of domain has installed the virtual machine server of different manufacturers, the virtual machine server of each producer belongs to the administration module management of this producer, the administration module being responsible for the management of cluster of domain load balancing is main management module, remaining administration module is from administration module, wherein, the mode of operation of the administration module of arbitrary producer all can be set to main management pattern or from management mode, described main management pattern refers to the load balancing management of responsible cluster of domain, describedly to refer in this management domain responsible the resources of virtual machine information of virtual machine server and the collection of virtual machine load information from management mode and report, and be responsible for sharing issuing of task, the administration module worked under main management pattern is main management module, and the administration module worked in from management mode is from administration module.Wherein, resources of virtual machine information comprises one or more of following information: the online information of virtual machine server, the resource behaviour in service information of virtual machine server; Wherein, the resource behaviour in service of virtual machine server comprises: the behaviour in service of CPU, the behaviour in service of internal memory.
Concrete grammar comprises the following steps:
Step 101. collects resources of virtual machine information in this administration module management domain and virtual machine load information by communicating with the virtual machine server that this administration module manages, from the resources of virtual machine information in administration module management domain and virtual machine load information described in reception reports from administration module, add up the resources of virtual machine information in described virtual machine server cluster of domain and virtual machine load information;
The inquiry of step 102. virtual machine server is in upstate and is externally flexibly connected the minimum virtual machine VM of number; If only inquire a virtual machine VM to be in upstate and to be externally flexibly connected number minimum, then using this virtual machine VM as target virtual machine VM;
If step 103. inquires and is in upstate more than the virtual machine VM of and to be externally flexibly connected number minimum, then in this multiple stage virtual machine VM, choose the minimum virtual machine VM of real-time traffic as target virtual machine VM;
Step 104. is on connection establishment to target virtual machine VM, respective virtual machine server is directly handed down to by this target virtual machine VM, as produced the load balancing task of virtual machine server from administration module management domain, then send to and issue from administration module accordingly.
In described step c, during virtual machine server inquiry virtual machine VM, if when a certain virtual machine VM is in and carries out the transition state moved between different server, the virtual machine VM that this is in transition state by virtual machine server is set to down state; After migration completes, again this virtual machine VM is set to upstate, virtual machine server adds up the external flexible connection number of this virtual machine VM, makes this virtual machine VM again participate in load balance scheduling.
Virtual machine server refreshes the external flexible connection number of target virtual machine VM; After refreshing, the external flexible connection number of this target virtual machine VM is: former external flexible connection number+1.
In the present invention, using virtual machine server as load equalizer, virtual machine server is connected with multiple servers, and each server internal is virtual turns to multiple VM.Each VM is the virtual machine independently externally providing service, externally connects.Because VM carries out exchanges data by load equalizer (i.e. virtual machine server), so load equalizer has flexible connection number statistics and the real-time traffic statistics of VM, therefore based upon activities linking number and real-time traffic carry out load balancing is feasible program.Load equalizer receive new outside time, first search and be flexibly connected the minimum VM (VM that namely idling-resource is maximum) of number, VM resource is used more balanced.After one period of scheduling time, the flexible connection number of each VM is tending towards identical, at this moment add up (outgoing direction of nearest a minute and the flow sum entering direction) according to the real-time traffic of VM again and carry out scheduling selection, by the scheduling of this combined strategy, external request can be evenly distributed on each VM.
Virtual machine VM has with external request and is connected, and also has service connection between virtual machine VM simultaneously.
Virtual machine environment is different from the environment of common server, because also have information interaction between VM, so will distinguish the connection between the connection of external request and VM.The flexible connection of load balancing foundation refers to the connection of external activity request.Because each VM is in the same network segment, and the address of external request is public network address or not at the same network segment, carries out differentiation external request and internal request by outgoing direction destination address.
In a virtual machine environment, VM can carry out " migration " between different server.Because load equalizer participates in this transition process, in transition process, this vm health is set to " unavailable ", after having moved, recover state and the linking number statistics of this VM, refreshing is transmitted, and again participates in load balance scheduling.During virtual machine server select target virtual machine VM, only select the VM being in upstate.
Although illustrate and describe embodiments of the invention, for the ordinary skill in the art, be appreciated that and can carry out multiple change, amendment, replacement and modification to these embodiments without departing from the principles and spirit of the present invention, scope of the present invention is by claims and equivalents thereof.

Claims (2)

1. the method for a virtual machine load balance process, be applied in the main management module of virtual machine server cluster of domain, described virtual machine server cluster of domain has installed the virtual machine server of different manufacturers, the virtual machine server of each producer belongs to the administration module management of this producer, the administration module being responsible for the management of cluster of domain load balancing is main management module, remaining administration module is from administration module, it is characterized in that comprising the following steps:
A. the resources of virtual machine information in this administration module management domain and virtual machine load information is collected by communicating with the virtual machine server of main management module management, from the resources of virtual machine information in administration module management domain and virtual machine load information described in reception reports from administration module, add up the resources of virtual machine information in described virtual machine server cluster of domain and virtual machine load information;
B. virtual machine server inquiry is in upstate and is externally flexibly connected the minimum virtual machine VM of number; If only inquire a virtual machine VM to be in upstate and to be externally flexibly connected number minimum, then using this virtual machine VM as target virtual machine VM;
Be in upstate more than the virtual machine VM of and to be externally flexibly connected number minimum if c. inquired, then in this multiple stage virtual machine VM, choose the minimum virtual machine VM of real-time traffic as target virtual machine VM;
D. on connection establishment to target virtual machine VM, respective virtual machine server is directly handed down to by this target virtual machine VM, as produced the load balancing task of virtual machine server from administration module management domain, then send to and issue from administration module accordingly.
2. the method for a kind of virtual machine load balance process according to claim 1, it is characterized in that: in described step c, during virtual machine server inquiry virtual machine VM, if when a certain virtual machine VM is in and carries out the transition state moved between different server, the virtual machine VM that this is in transition state by virtual machine server is set to down state; After migration completes, again this virtual machine VM is set to upstate, virtual machine server adds up the external flexible connection number of this virtual machine VM, makes this virtual machine VM again participate in load balance scheduling.
CN201410077897.8A 2014-03-05 2014-03-05 Virtual machine load balance processing method Pending CN104899094A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN201410077897.8A CN104899094A (en) 2014-03-05 2014-03-05 Virtual machine load balance processing method

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201410077897.8A CN104899094A (en) 2014-03-05 2014-03-05 Virtual machine load balance processing method

Publications (1)

Publication Number Publication Date
CN104899094A true CN104899094A (en) 2015-09-09

Family

ID=54031772

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201410077897.8A Pending CN104899094A (en) 2014-03-05 2014-03-05 Virtual machine load balance processing method

Country Status (1)

Country Link
CN (1) CN104899094A (en)

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
WO2018184302A1 (en) * 2017-04-06 2018-10-11 平安科技(深圳)有限公司 Data transmission method, system, virtual load balancer and readable storage medium
CN112099835A (en) * 2020-09-07 2020-12-18 紫光云(南京)数字技术有限公司 Soft load balancing mirror image dynamic upgrading method and system

Cited By (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
WO2018184302A1 (en) * 2017-04-06 2018-10-11 平安科技(深圳)有限公司 Data transmission method, system, virtual load balancer and readable storage medium
US10693785B2 (en) 2017-04-06 2020-06-23 Ping An Technology (Shenzhen) Co., Ltd. Method and system for forwarding data, virtual load balancer, and readable storage medium
CN112099835A (en) * 2020-09-07 2020-12-18 紫光云(南京)数字技术有限公司 Soft load balancing mirror image dynamic upgrading method and system
CN112099835B (en) * 2020-09-07 2022-10-28 紫光云(南京)数字技术有限公司 Soft load balancing mirror image dynamic upgrading method and system

Similar Documents

Publication Publication Date Title
US8949847B2 (en) Apparatus and method for managing resources in cluster computing environment
CN105183554B (en) High-performance calculation and cloud computing hybrid system and its method for managing resource
CN102929718B (en) Distributed GPU (graphics processing unit) computer system based on task scheduling
CN105207798B (en) Service arrangement method and device in software defined network
CN102650950B (en) Platform architecture supporting multi-GPU (Graphics Processing Unit) virtualization and work method of platform architecture
CN103731295B (en) Method and system for running virtual integration equipment
CN102594881B (en) A kind of virtual machine load-balancing method, management module and virtual machine cluster system
CN102271145A (en) Virtual computer cluster and enforcement method thereof
CN103401947A (en) Method and device for allocating tasks to multiple servers
CN202918339U (en) Ground test-launch-control system of carrier rocket based on cloud computing
CN102110014A (en) Method for balancing loads of virtual machine (VM)
CN104038540A (en) Method and system for automatically selecting application proxy server
Zhang et al. A statistical based resource allocation scheme in cloud
CN104050043A (en) Share cache perception-based virtual machine scheduling method and device
Patni et al. Load balancing strategies for grid computing
CN103036800A (en) Virtual machine load balancing system, balancing panel points and balancing method
CN106293933A (en) A kind of cluster resource configuration supporting much data Computational frames and dispatching method
CN102929769A (en) Virtual machine internal-data acquisition method based on agency service
CN106020969A (en) High-performance cloud computing hybrid computing system and method
CN106059940A (en) Flow control method and device
CN106961440B (en) Cloud platform based on the operation monitoring management of enterprise-level resource
Shen et al. Facilitating application-aware bandwidth allocation in the cloud with one-step-ahead traffic information
CN104899094A (en) Virtual machine load balance processing method
CN105681424A (en) Desktop cloud system
Chen et al. A live migration algorithm for virtual machine in a cloud computing environment

Legal Events

Date Code Title Description
C06 Publication
PB01 Publication
WD01 Invention patent application deemed withdrawn after publication

Application publication date: 20150909

WD01 Invention patent application deemed withdrawn after publication