CN102681899B - Virtual computing resource dynamic management system of cloud computing service platform - Google Patents

Abstract

Cloud computing is the third change of information technology (IT), and deeply affects information acquiring modes of the information-based society. A cloud computing service platform converts computing, network and storage into a resource pool to provide service for a user through the network. According to a virtual computing resource dynamic management technology of the cloud computing service platform, a layered cloud computing infrastructure model is provided; three kinds of objects are defined to describe and manage statuses, attributes, characteristics and behaviors of a telescopic virtual machine resource; the supply from a physical server to the virtual resource pool is automated; and an online dynamical capacity-expanding method is provided, and through real-time acquisition of load status and application running conditions of the system, the virtual computing resource is dynamically expanded under the condition that the service is not interrupted by taking the artificial neural network as a mechanism for judging expansion and triggering of the resource.

Description

The virtual computing resource dynamic management method of cloud computing service platform

One. technical field

The third time that cloud computing is considered to after microcomputer, internet in IT field is changed, and is the trend of the times of internet development.

Cloud computing is the development along with Intel Virtualization Technology, CPU multi-core technology, broadband technology, concurrent operation, Distributed Calculation, grid computing, the continuous expansion of internet scale, the demand of user to information, network, calculating, storage constantly increases, the technology that service provider is this demand of adaptation and grows up gradually and service.

Cloud computing by using as IT information service provider the shielding of the ins and outs such as server, operating system, network, storage, basic software, platform, framework and abstract be the service form such as computing power, storage capacity of management, dynamically be supplied to information user by network, user pays by actual use amount.

Cloud computing has following key point: the resource 1) shared and the Technical Architecture 2 shared) there is distributed computation schema and memory module; 3) statistic multiplexing of data center's hardware resource and flowable resource is realized by virtual; 4) the IT resource service of the different stage comprising software and hardware facilities can be provided for user; 5) resource dynamic expansion, supplies, according to quantity charging as required.6) intelligentizedly Resourse Distribute is applied to; 7) application deployment of robotization and resource distribution.

Present cloud computing has been not merely a kind of technical term or Technical Architecture, especially a kind of new service mode and business model.

Cloud computing is by integrating the server of technical field, network, storage, various operating system, software; and further the data center in the company in industrial field, unit and enterprise, equipment, personnel, service etc. are integrated; information service is made to accomplish industrialization, scale, specialization, intensive; greatly improve the utilization rate of Information base resource and the efficiency of information service, save social cost, resource and consumption.

Two. background technology

The core technology of cloud computing contains Intel Virtualization Technology, distributed computing technology, cloud platform management technology

2.1 Intel Virtualization Technology

Virtual development of increasing income based on Xen since 2006 is swift and violent, Intel Virtualization Technology is progressively started extensively universal.Because Intel Virtualization Technology can install multiple operating system on a station server, thus substantially increase the service efficiency of server.

The virtual logical expressions representing resource, in theory not by the restriction of physical resource.The way of realization of Intel Virtualization Technology adds a virtualization layer in an operating system, the Resource Abstract of lower floor become the resource of another form, is supplied to upper strata and uses.Have that the virtual and T/O's of virtual, the internal memory of CPU is virtual in practice.

The technique although IBM just began one's study as far back as the seventies, but realize on X86-based virtual from VMware the nineties, because X86 is difficult to realize virtual, this is because the ISA framework of X86 has 17 sensitive instructions not belong to privileged instruction, if that is virtual machine performs the instability that these instructions agree cause whole system.

So Vmware is proposed fully virtualized commercial version product, namely without the need to revising the operating system of client computer, by intercept and capture client computer perform privileged instruction and scan client computer kernel perform binary code, sensitive instructions is translated into virtualization instructions to perform.

Within 2005, there is half virtualization product of increasing income taking Xen as representative later, by the code of amendment Guest OS, make its by those operations relevant with privileged instruction all conversion can issue the Hypercall (hypercalls) of VMM, and Hypercall supports batch processing and these two kinds of optimal way asynchronous, makes the speed that can obtain being similar to physical machine by Hypercall.

After 2007, Intel and AMD is proposed the virtualized product of support hardware, by introducing new cpu instruction and operational mode, virtual machine and VMM is allowed to run on different processor modes, as Root pattern and Operation pattern, when virtual machine performs sensitive instructions, system automatically switches to Root pattern and is performed by VMM.

Also having a class virtual is the operating system virtualization product that is representative with OpenVZ and Solaris Container, and by the operating system partition of main frame, the basis of kernel execution instance realizes virtual, system overhead is less.But owing to not providing virtual operating system environment, manyly need to access the application of bottom layer driving and cannot run, and the operating system different from host operating system also cannot be run.

Although Vmware provides many effective virtual machine management systems, but because VMware is business version, version expense is collected according to the CPU slot number on main frame, need like this to rely on magnanimity PC server to provide the user of service for Internet firm, the cost using VMware is high.

And based on the such virtualization product of increasing income of XEN or KVM, although also there is the dynamic migration function of VM, but lack the management system of virtual machine, when tackling magnanimity resources of virtual machine, lack effective management tool, automatically cannot distribute resources of virtual machine and supply resources of virtual machine, also shortage calls resources of virtual machine with the interface of application is application service.

2.2 distributed computing technology

Distributed treatment is a kind of mode of information processing, and be a concept relative with centralized processing, the multiple stage computing machine being dispersed in various places is coupled together by communication network by it, under the management of control system controls, completes information handling task in phase.Distributed treatment is usually used in carrying out analytical calculation to mass data, it is computing machines different in data and distribution of computation tasks to network, these computing machines jointly complete calculation task under the scheduling of controller, and the performance of distributed treatment depends primarily on the communication efficiency of data and control.

Distributed treatment is a key link of cloud computing, it can be deployed in virtual on, solve the collaborative work problem of cloud computation data center large-scale server group, be made up of distributed file system, Distributed Calculation, distributed data base and distributed synchronization mechanism four parts.

Distributed file system is one of architecture of distributed computing environment, it presents to user dispersion file resource in a network with unified viewpoint, simplify the complicacy of user's access, strengthen the manageability of compartment system, also prepared condition for further development DDB application.Distributed file system is based upon on Client/Server Technology basis, by server and client file systems co-operating.Controlling functions dispersion between client and server, makes the thing such as shared, data security, the transparency etc. are easy to process in centralized file system become quite complicated.File-sharing can be divided into read to share, sequential write shares and concurrently writes shared, in distributed file system, sequential write needs the same viewpoint problem solving sharing users, concurrently writes, and needs to consider middlely to insert the consistency problem upgrading and cause.In data security, need date restoring during private ownership and the conflict considering data.Transparency demand file system is unified complete to the interface of user, at least needs to ensure the transparent and failure transparency of location transparency, Concurrency Access.In addition, extendability is also the problem that distributed file system needs emphasis to consider, when increasing or reduce server, distributed file system should be able to automatic sensing, and does not have any impact to user.The example of distributed file system has GFS, HDFS, Ceph, Lusture, glusterFS, NFS etc.

Distributed data base is one group of structurized data set, belongs to same system in logic, and on the multiple places physically disperseing active computer network to connect, and unified by a distributed data base management system (DDBMS) management.Compared with centralized or separate data storehouse, distributed data base has the advantages such as reliability is high, module expansion is easy, operating lag is little, load balancing, fault-tolerant ability are strong.Data redundancy is one of distributed data base principal character being different from other databases, which ensures the reliability of distributed data base, is also parallel basis.Along with the development of web2.0 dynamic website taking SNS as representative, high concurrent, mass memory and height have been extended to the bottleneck of traditional Relational DataBase, and the non-relational database NoSQL addressed these problems grows up gradually.The example of distributed data and NoSQL database has BigTable in a word, HBase, MongoDB, Redis, Tokyo Cabinet, CouchDB, Cassandra etc.

Distributed Calculation be allow several physically independently assembly as an independent system synergistic working.For distributed programmed, the problem of core be how a large application program resolve into some can the subroutine of parallel processing.Have two kinds can treatable method, one is separation calculation, namely the function of application program is divided into several modules, has been worked in coordination with by multiple stage machine on network; Another kind is partition data, namely Segmentation of Data Set is become fritter, is calculated respectively by the multiple stage computing machine on network.Large-scale distributed system can take these two kinds of methods usually simultaneously, and solves the problem of collaborative work between each functional module.This type systematic may adopt based on C/S structure three layers or multilayer distributed Object architecture, and presentation logic, service logic and mathematical logic are distributed on different machines, and CORBA, EJB, DCOM are the middleware Technology of three kinds of main flows.Also may adopt the architecture of sing on web, or claim Web Service, this model is the Enterprise SOA of highly distribution, and SOAP, WSDL, UDDI are the core technologies of Web Service.

In a distributed system, lost update may be caused to the parallel work-flow of shared resource, read the data such as dirty data, non-repeatable read inconsistence problems, at this moment need to introduce distributed synchronization mechanism to control concurrent operations, the most frequently used mode is distributed lock mechanism and collision detection.There is several conventional concurrency control method:

Based on the concurrency control method of lock mechanism

Based on the concurrency control method of timestamp

Optimistic Concurrency controls (Optimistic Concurrency Control) method

Based on the concurrency control method of version

Based on the concurrency control method of transactions classes

For the cloud computation data center be made up of extensive low-cost server group, distributed synchronization mechanism is the basis of carrying out all upper layer application, is the basic guarantee of system correctness and reliability.Google Chubby and Hadoop ZooKeeper is the Typical Representative of cloud architecture distributed synchronization mechanism, and for each parts of coherent system, other distributed systems can carry out synchronization of access shared resource with it.

2.3 cloud platform management technology

The major function of cloud computing is effectively managed computational resource, storage resources, Internet resources, is supplied to user in the form of services.So singly do not need Intel Virtualization Technology and distributed computing technology as basic fundamental element, the service platform that user calls computational resource is supplied to as one, more need the integration processing virtual resource and physical resource, solve resource demand assigned problem, the problem of dynamic expansion, the problem of automatic management, the problem of application deployment, the problem of data management, and the problem such as user management, carrying out safety backup, this cloud platform management system that just needs one are perfect has come.

The technology such as the Intel Virtualization Technology more than related to, Distributed Calculation, distributed storage are all the basic fundamental assemblies of cloud computing, but large-scale network system, database, basic software etc. to be supplied to user in the form of services as computing power, storage capacity etc., also need the administrative skill of cloud platform by these component organization and management, realize the deployment of robotization, dynamic expansion, the support of application, the access of data etc.

2.3.1 the service mode of cloud platform and COS

And cloud computing is a kind of service mode of innovation, and the service mode of cloud platform can be divided into IaaS, PaaS and SaaS tri-kinds according to according to technical system framework:

IaaS

The bottom of cloud computing platform is IT infrastructure, Infrastructure as a Service be exactly service provider IT infrastructure is comprised server, network, storage, operating system virtual after become the manageable virtual resource of a kind of program, by network for user provides calculating and stores service.Typical example such as the EC2 of Amazon, Gogrid and Rackspace are also cloud computing IaaS service providers.

PaaS

PaaS platform is the platform of a kind of exploitation and trustship network application, this platform provides several network programming language and framework, user can use this programming language and framework to write easily extensible and flexible application, runs in cloud computing platform, and without the need to considering whether program is subject to the restriction of resource.This is generally that service provider designs the application call that a kind of programming framework API allows user write and operates on the cloud computing facility of oneself inside.The Apps Engine of such as Google.

SaaS

Namely software serve, such as Google Gmail, Google Docs.This is a kind of pattern being provided software service by internet, and manufacturer is by application software unified plan on the server of oneself, and user does not need to download and install this locality, can be used by browser.User can according to oneself actual demand, and by the service of ordering, how many and time length is to manufacturer's defrayment.User need not buy software again, and without the need to safeguarding software, service provider understands full powers and administers and maintains software, software vendor is while providing internet, applications to client, off-line operation and the local datastore of software are also provided, allow user can use its software and services of ordering whenever and wherever possible.

Cloud computing platform can be divided into again privately owned cloud and publicly-owned cloud according to COS.Dispose within enterprises and fire wall, for enterprises IT with to produce what provide service be privately owned cloud; Dispose beyond fire wall, what provide public service for the public is publicly-owned cloud; For cloud computing client, oneself have privately owned cloud, again a part of data and calculating are put into publicly-owned cloud platform, the using forestland of this cloud computing is mixed cloud simultaneously.

2.3.2 the administrative skill analysis of cloud platform

For IaaS, the management of cloud platform to be concentrated by unused physical resource with Intel Virtualization Technology and after management gets up, considers how these abstract virtual resources are supplied to user, and therefrom create economic benefit.Need manage IT infrastructure on the one hand, comprise shielding hardware differences, monitors physical resource using status, dynamic assignment virtual resource etc.; Also need to provide the interface with user interactions on the other hand, comprise the api interface that standard is provided, the configuration interface that virtual resource is provided, provide service catalogue to search available service for user, provide real time monitoring and statistical function etc.

PaaS is served, cloud platform needs dispose the technology such as distributed storage, distributed data base, distributed synchronization mechanism and distributed computing model, after making platform just possess the basic capacity of distributed software development, PaaS cloud service provider also needs to consider how this development platform is supplied to user, as in user interface, need to provide code library, programming model, DLL (dynamic link library), development environment etc.The storage of code library package platforms, calculating, database basic function, during for User Exploitation application program.Programming model determines the Application Type of user based on cloud platform development, and it depends on the distributed computing platform of platform selecting.The supply of the storage needed for PaaS operation management system need solve in user application operation process, calculating, network foundation resource and problem of management, need dynamically increase according to application program practical operation situation or reduce running example.For ensureing the reliability service of application program, system also needs to consider the mutually isolated problem between different application, allows their reliability services in the sandbox environment of safety.

For SaaS, build on the internet owing to serving itself, user possesses networked capabilities and can use online.Regardless of the operation management system of any service, all need system management problems and the user management such as charging, certification, safety, monitoring that solution product relates in operation process.In addition, for the difference of business characteristic, each service operation management system also needs to solve problems different separately.

SaaS pattern was just present in before cloud computing occurs, but SaaS pattern does not also open up new prospect all the time in China.A lot of SaaS company has all closed down, and also only has Salesforce mono-more successful in the U.S..

The main flow of existing cloud computing service platform is exactly PaaS product and IaaS two type.

PaaS service is for representative with Google and Microsoft, main employing concurrent operation, distributed arithmetic, distributed storage technology, the client be suitable for has large-scale processor active task, as data mining, data analysis, search etc., user needs a kind of new multiple programming method of study or programming tool, and use this new tool to carry out again coding to complete calculation task, though user can shorten the operation time of application exponentially, but original application can not be transitioned into new system smoothly.Such feature the application of user is broken up to rub broken being then distributed in resource pool, sends to each to be in the server process of diverse geographic location calculating, obtain result and return as task.

IaaS service is for representative with Amazon and RackSpace, main employing Intel Virtualization Technology, being applicable to single calculation task is not super large, but single application is monopolized server and wastes computational resource, simultaneously towards numerous such application and calculation task, or towards numerous such client.Such service, by hardware server resource virtualizing subregion, is supplied to multiple calculation task or multiple user service, and manages the resource pool of numerous virtualized server formations.For user; original application can move in the environment of cloud; and do not need the new programming tool of study again to write application; the application program of installing other is not needed yet; thus original investment can be protected, such feature is that the resource in resource pool is dynamically assigned to the application needing resource as required.

At present commercially, the cloud computing service of IaaS account for the overwhelming majority.This is because market is relatively ripe, business model is relatively clear, and the application of user more easily moves in cloud computing environment, and the first pattern is more suitable for developer in cloud environment, directly writes the application of new cloud.

IaaS service is the cloud computing mode constructed in architecture, need the virtualization product of bottom, as VMware, Xen, KVM etc., wherein Xen/KVM is the product of increasing income, Redhat is integrated with Xen and KVM, these softwares are called as HypeVisor or VMM, achieve the virtual machine resource management in separate unit physical machine, comprise CPU, internal memory, the virtual management of I/O and Virtual Machine Manager.But how numerous virtualized servers and network, storage etc. are managed as a resource pool, form a cloud computing platform, intelligent automaticization ground is for applying the resource providing dynamic on-demand to distribute, independent virtualization software is insurmountable, this needs the management software of a cloud computing, or is called cloud platform management system and has come, and this respect does not have standard, do not have ripe product and scheme, rare successful case can be used for reference.

As seen from the above analysis, PaaS platform service is the most complicated, both large-scale data center and network system platform had been needed to support, need again to provide perfect distributed application program development environment, DLL (dynamic link library), code library, programming model, distributed data base etc., only have Google and Microsoft really to have the ability to provide with Apps Engine the PaaS being core to serve at present.Domestic also only have Sina to test SAE.

PaaS platform service to as if application developer, audient face is less, and developer needs, from the new multiple programming mode of new study, more to increase and use difficulty.

The business model discomfort of PaaS is very clear, and the COS for Apps Engine is said, be difficult to realize, and due to the developer towards minority, real income is not high yet according to request call amount, the charging of CPU use amount application programs.

Cloud computing operation is real successfully, that have clear profit model is IaaS, and take Amazon as representative, many companies such as RackSpace, GoGrid are this pattern of self-replication all.And most users is using EC2 and the S3 service of Amazon on current cloud computing service market, such as maximum in the world video display lease service business Netflix made the transition in the mode of video flowing for user provides VOD and download service, and they also from original self-built from operation/maintenance data center to the cloud service used based on Amazon EC2 because EC2 can provide linear expansion.

2.3.3 IaaS platform management technology

Amazon EC2 is that typical IaaS serves, providing can the virtual machine of installing operating system, user can oneself set up applications, also can use AMI (Amazon Machine Image), can build the customized application of a high complexity with Amazon EC2.EC2 allows user to control environment parameter, underlying operating system, storage and network demand, and technically, it belongs to the service of very bottom, and user can adjust required most of thing.

The EC2 of Amazon provides a platform as IaaS service, and its service is to provide calling of resources of virtual machine and configuration service, but calling interface is a series of api routine interfaces of Amazon oneself definition.Many third-party platform developers can develop third-party application on Amazon platform, and the robotization realizing user application is disposed, the dynamic expansion etc. of resource.If user selects from service, then need the api routine interface learning Amazon EC2 service, then oneself compile script application and create virtual machine environment, the deployment of application program also needs user oneself to complete.

Joyent, Gogrid is also had, the CloudEx etc. of Rackspace, Savvis, Terremark and A-1. Net with the similar IaaS cloud platform of Amazon EC2.Owing to lacking unified interface and communication standard, the interoperability between these platforms is very poor, user cannot between IaaS service provider seamless migration.Abiquo company is proposed the cloud computing platform AbiCloud that increases income, The platform provides unified cloud core, can be used for creating the publicly-owned cloud and privately owned cloud with extended capability, additionally provide the various interfaces for integrating third-party instrument and software, the construction of assistance services provider meets the various clouds of particular demands, attempt in IaaS standardization and the effort made in increasing income, but these attempt being not sufficient to ensure that IaaS operation management system is available conveniently.Service provider needs as required, selects research and development voluntarily or on the basis of existing open source projects, builds the IaaS operation management system meeting enterprise's particular demands.

The basic function that these IaaS platforms itself realize is that a better simply virtual machine distributes, even if but the system of its inside realizes also very complicated like this, what Amazon was never disclosed its background system realizes details, but some open source software claims the repertoire achieving Amazon EC2, such as OpenNebula, Nimbus, OpenStack, Eucaptus.

In a word, the management of cloud computing service platform is a complicated problem, and current industry does not also form relevant standard, does not also have to bring the system of directly disposing and using, and cloud service provider needs to realize separately.

2.3.4 increase income product and the problem thereof of IaaS platform management system:

Xen Cloud Platform

Xen Cloud Platform is the project initiated by the developer community of the hypervisor Xen that increases income, be devoted to be supplied to the complete Infrastructure platform of service provider one, possess open and measured API, multi-user leases, service level agreement ensures, the feature of charging as required.This project originates in 09 year, and be at present in fast Development, system is not mature enough, and Install and configure is comparatively complicated.

Eucalyptus

Eucalyptus is the software for realizing cloud computing part infrastructure, allows user make full use of the server zone of oneself.

Eucalyptus only realizes basic simple management function of virtual machine, and its main target is not the powerful of management function, but by basic virtual management APIization.It has following characteristic: with EC2 interface compatibility (primary interface) Rocks cluster management instrument, install and dispose all very simple and safe intercommunication, use the working method of SOAP and WS-Security superposing type, without the need to " cloud management " instrument that modifying target Linux environment is basic, realize system and nusrmgr.cpl, multiple cluster configuration can be become single cloud.

Nimbus

Nimbus is the cloud computing project of increasing income under grid middleware Globus, and Nimbus, towards scientific algorithm demand, carrys out by one group of Open-Source Tools the cloud computing solution that namely optimized integration facility serves (IaaS).In Nimbus platform, the assembly comprised has: Workspace node manager, realize based on the remote protocol of WSRF, realize based on the remote protocol of EC2, cloud computing client, WorkspacePilot integrate virtual machine etc. towards the application component of different aspects, each assembly of Nimbus project is very lightweight and possess self completeness in design, can be combined by multiple isomery mode.

OpenNebula

OpenNebula is that construction is privately owned, public, the Open-Source Tools of mixed cloud.Framework, interface, assembly are flexibly provided.Key property comprises: support Xen, Kvm, Vmware, virtual platform can dock EC2 and ElasticHost, uses libvirt, supports EC2 and OGC OCCI interface

Libvirt

Libvirt is the C function library of the virtual instrument of main flow under a set of free, support Linux of increasing income, and its various virtual instrument be intended to for comprising Xen provides a set of convenience, reliably DLL (dynamic link library), supports the binding with the multiple main flow development languages such as C.The virtualization management tool virt-manager (graphically) that current main-stream linux platform is given tacit consent to, virt-install (command mode) etc. all form based on libvirt exploitation.Feature: primary C language interface, and the binding interface of other active languages, meet CIM and QMF specification, supports Xen, Qemu, KVM, LXC, OpenVZ, UML, Virtualbox, VMWareESX, can telemanagement virtual machine, virtual network.

These open source softwares achieve most of function of Amazon EC2 really, then the distribution of its essence or computational resource and storage resources, certainly the service of user, the management of user and certification etc., and the function such as automatic management, the dynamic expansion of resource, the robotization deployment of application really do not realized required by cloud computing platform, at Amazon platform, these all rely on third-party platform developer to provide, as RightScale etc., but the software of these companies is all business-like.User can not freely obtain.

2.3.5 the commercially produced product of IaaS platform management system and problem thereof:

The business version of cloud plateform system management software has: VMware vSphere/vCloud, Dell VIS, Platform ISF, Heroku, RightScale, BMC, IBM, HP, OpSource etc.

VMware

VMware is used for product mainly vSphere and vCloud that be virtual and cloud computing platform management.

It is virtual that vSphere is mainly used in server end, plays the object such as Server Consolidation and resource optimization by fictionalizing multiple stage virtual machine on a physical server.VSphere can be two parts: first VMM (virtualization manager Hypervisor) part, it two is vCenter for integrating and manage VMM, major function has resource and virtual machine inventory management, task scheduling, log management, warning and practice management, deploying virtual machine and arranging.

VCloud Director is the cloud computing plan of VMware, You Liangge branch, and the first is called the IaaS solution of VMware vCloud Express, and it two is PaaS solutions of VMware Platform as a Service by name.

VCloud Director is the virtualization capability based on VMware vSphere, and the resource pool function extending VMware vCenter creates " VDC (Virtual Data Center; virtual data center) " to enable IT department, namely by calculating, resource pool that network and storage resources forms and predefined operating strategy, service level agreement and pricing mechanism, and for user provide based on VDC computational resource with can application deployment on it.VCloud is support resource isolation and many tenants in design, and vCloud introduces the concept of two unusual cores for this reason: the VDC of the first for isolating resource; It two is tissues (organization) for supporting many tenants mechanism.

VDC is a set comprising for resources such as the calculating of cloud computing and storages, and multiple user is combined into same tissue by rule (Policy) by keeper.

The businessman that on current cloud platform management system market, share is maximum is exactly VMware; but its product line is a lot; in order to protect the virtualization product before it; its the new product towards cloud computation data center all increases on existing product; platform product and data center products are also separated, and the division of the isolation of resource and tissue all needs manual operation, and function can be very complicated; need special VMware managerial personnel could manage, this both increases IT handling cost.

Dell VIS

Dell VIS framework can dynamic adjustments application load and by calculating, store and networked asset be integrated into unified resource pool.VIS framework is made up of modular assembly, can be mutually integrated with client existing IT environment.VIS framework comprises following three primary clusterings:

Senior infrastructure manager (AIM), support that keeper is according to application load distribution server, storage and Internet resources, AIM can extract hardware and virtualization layer from data center, and therefore, client can be absorbed in configuration single resource pool and the multiple different technology of non-management.

The self-service creator of VIS, can standardization automatically perform the deployment flow process of application, and then the deployment time of operating load is foreshortened to a few minutes.The self-service creator of VIS can strengthen IT control, accelerates IT flow process simultaneously, thus saves time and resource.

Dell VIS Director, supports user to check virtual dependence comprehensively and identifies the problem in virtual environment fast.This module comprises advanced report, suppose and trend analysis, capacity and utilization rate report, absorption of costs and refuse to pay solution.

Dell is also the important producer of cloud computing platform management software field one, and VIS can be responsible in dynamic adjustments application, but its hardware automated deployment can only run on the hardware platform of Dell oneself, does not support the hardware platform of isomery.

Platform ISF

Platform is also proposed cloud computing platform management product ISF in the recent period, it is the Automation Construction for the privately owned cloud of enterprises, can integrated various application environment as J2EE, webSphere, Test/Dev etc., and have multiple resources allocation strategy, key business can reserve resource, and product itself does very well.But Platform has done scientific algorithm HPC field since the more than ten years in the past, just forward field of cloud calculation to recently, so his product nature is with the vestige of scientific algorithm, be assigned in different virtual machines as controlled application load, this can substitute with the equilibrium of LSF even load in fact.

More multiaspect is to enterprises large-scale application for this product, instead of internet web applies.

The price of ISF is very expensive in addition, collects the version expense of $ 795 U.S. dollar according to each CPU slot of server.For Internet firm, mostly adopt the PC server of magnanimity cheapness, this price is that no one can bear.

BMC

BMC releases a cloud computing life cycle management scheme, comprises the service catalogue that strategy promotes, and to customize service, for the Self-Service of cloud computing resources request and control; Be suitable for the architecture of privately owned cloud and public cloud service, for comprising calculating, network, store, be applied in interior whole service stack and carry out dynamic-configuration; Based on secure web services and the cloud management work stream namely of grading performance; The performance of performance monitoring instrument supervision cloud service; Based on the realization flow of ITIL.

Cloud computing life cycle management scheme is under the jurisdiction of BMC cloud service Managed Solution series, the cloud platform set up by the program, can realize the robotization to many tenants cloud architecture physics and virtual dividing, load balance, fire wall and service level management can be connected on each container automatically; Can realize " Self-Service " cloud managing portal, and simplify the operation, Seamless integration-is to the planning of public and privately owned cloud architecture, allotment and management.

But BMC does software operation, and flow process is built up, so his cloud computing management software lays particular emphasis on business service workflow management more, the management of cloud computing that what the boss of BMC also said that BMC does is, that does is not to provide cloud computing, but first must to have and integrate physics and virtual computing resource, the service of cloud computing just can be had to provide.So the software of BMC is the client for employing as AmazonEC2 cloud computing, namely the Amazon EC2 of use is managed with the cloud management software of BMC, so this cover software is for will developing and managing cloud computing infrastructure and providing the user of cloud service.

Domestic from research and development: Huawei, A-1. Net, transports greatly, tide

Current domestic a lot of IT giant sets foot in cloud computing industry to a high-profile as Huawei, association, tide, Founder, dawn, Alibaba, A-1. Net etc. are also numerous and confused, drop into huge fund to attempt to occupy a tiny space in the cloud computing industry of China, a lot of company separates even separately whole family company and specially makes this business.

Huawei develops a set of cloud computing platform management software on the basis of open source software Eucalyptus, is just attempting now the large operators in co-operation with China Mobile, UNICOM and telecommunications three, is releasing their cloud computing software and hardware solution.Tide and dawn etc., also in exploitation cloud operating system, are contained virtual, resource management, task scheduling, load balancing, task management, safety certification, charging etc., but are not also commercially released.

What market can be seen is the cloud cable release of A-1. Net, at present this product or semi-automatic, also very immature in self-help service for user, resources configuration management, dynamic expansion.

Cloud computing platform management system is done by open source software by most of company of current China, because this market is immature, client did not meet the product what is cloud computing yet, so be easy to the market obtaining the initial stage, but open source software also has a lot of problem, this several open source software is all the basic function that virtual machine can only be provided to distribute, and the resource based on strategy is distributed and resource dynamic expanded function automatically, does not also have the robotization supply of hardware platform.And these open source softwares have made the less functions used such as a lot of network pool and storage pool, but in actual applications, network pool can rely on LVS to realize, and storage pool also can lean on distributed file system to realize.

And open source software itself directly can not bring the through engineering approaches software doing enterprise operation, bug is a lot, and not through pressure test, lack the mechanism of managing risk and process problem, reliability is not high, cannot bear through engineering approaches task.The community itself that simultaneously increases income always can not provide technical support timely and help yet.

The system that such as Huawei commercially pushes away now is not just increase income completely based on Eucalyptus, has a lot of Premium Features such as load balancing and expansion just not to increase income.

Business software in the market and have such-and-such problem based on the software of increasing income in a word: lack automatic management, homogeney is high, function is too complicated, need too many human configuration, lack robotization supply, lack resource dynamic expansion, external commercialization cloud computing management software does not also Chinesize, and domestic software does not also develop.

Three. summary of the invention

This invention is the virtual computing money dynamic management system of a cloud computing service platform, solves the various problems of above-mentioned open source system and commercial system, can provide the supply of virtual resources, configuration, distribution, expansion, management.Be absorbed in the rapid deployment of cloud computing resources, supply, the layer-management of resource, the high effective integration of resource and elastic telescopic that is shared, resource, Dynamic trigger mechanism, the linear expansion of resource pool.

3.1 features:

Shared physical resource pond can be generated, share between different department and different application

The layer architecture model of corresponding service logic

Support the virtual drivings of main flow such as Xen, KVM, VMware, OpenVZ, adopt Libvirt to call

Support the virtual meter operating systems such as Windows, Solaris, Linux

Support the agreements such as Kickstart, SystemImager, TFTP, PXE, DHCP

Generate flowable virtual resource pond, eliminate Single Point of Faliure, increase stability

Intelligentized resource, to application allocation strategy, reduces human assistance.

The physical server of robotization, to the fast supply in virtual resource pond, enhances the agility of service response.

The application deployment of robotization

Resource dynamic is expanded, and can not affect the service of application program.

Adaptive dilatation trigger mechanism, adjustable activation threshold value.

Application program load balancing is integrated

Distributed buffer memory

Dissimilar application program is integrated, J2EE, HPC, Test/Dev

Administration interface is succinct, provides command line interface.

3.2 functions:

There is provided patterned friendly interface allow user oneself define or select needed for computational resource, the mode that user can pull with mouse builds an enterprise IT or internet, applications framework on cloud platform.

System can dynamically generate and pay the computational resource of meeting consumers' demand in several minutes.

The user's request of cloud platform management system automatic analysis, the strategy good according to predefine distributes resource requirement.

User without the need to Install and configure operating system, the system installation of complete operation system in all empty machines.

User without the need to oneself Install and configure application program, the application program automatic deployment that user submits to by management system, and load balancing service, rank management, Account Administration, fault-tolerant management, backup management etc. are provided.

Cloud plateform system provides the basic IT service such as buffer service, web services, mail service, directory service, VPN service for all users.

Cloud plateform system provides dynamic expansion service as required for user, the program that user writes without the need to considering resource expansion problem, the load pressure problem that mass users concurrency and flow bring.

User is without the need to learning new programming tool and programming framework, and original application program can seamlessly move on cloud platform.

Cloud plateform system can complete hardware server initialization and virtual automatically, and Auto-mounting configuring virtual machine, composition virtual resource pond.

System realizes automatic management, decreases manual operation.Only need when system is initially installed allocated resource hierarchical levels and system dynamic expansion strategy, in service afterwards, system can automatically be distributed and extended resources, and operator only need add physical server in resource pool.

System improves server operational efficiency and service efficiency in a word, enhances platform stabilization, improves automatic management ability, reduces the service response time.

3.3 principle of work:

The technological frame figure of a typical internet, applications as shown in Figure 1.Each square frame represents the specific function that batch processing completes.Square frame in each horizontal direction is on same service logic level, and the logic trend of whole application is from top to bottom by shown in the direction of arrow.Large-scale Internet enterprises or other IT enterprises are all have multiple same or similar like this application deployment on such framework, and the program assembly of each application binds specific server resource.

So according to the logic flow of business, this framework can be abstract in shown in Fig. 2, namely resource is divided into following different level.Server hardware resource is divided into multiple logical levels according to predefined strategy after virtual.

Each level is abstracted into the resource pool of a virtual machine composition.

The resources of virtual machine configuration of each level is not identical.

Virtual machine configurations all within level is identical.

Each level supports different business, and each business is the resources of virtual machine pond of this level shared in specific level.

Each business has the trend of logic business and data at longitudinal level.

Define 3 kinds of objects in systems in which, Aggregate, Vol, vTask.

Aggregate represents the polymerization of all similar virtual machines (VM) to liking one, the virtual machine set of the corresponding level as shown in Figure 3 of each Aggregate.

Vol, to the subset object liked on Aggregate, is represent the virtual machine set of a certain business in the distribution of this level.

VTask object represents the set of virtual business vol required on each level, contains service logic flow process.

Relation between three is as shown in Fig. 4 .3.

The feature of each object:

Aggr：

Aggr is the object of the polymerization representing a kind of identical type virtual machine, and each virtual machine in polymerization has identical attribute, such as identical CPU quantity and frequency, memory size, hard-disk capacity etc.

When system generates the example of Aggr, only need specify a series of physical machine name with the configuration of identical virtual machine, system distributes to the example of Aggr by the continuous VM in these physical machine.

The bottom of Aggr is physical machine.

Aggr can by adding machine extended by hands.

Vol

Vol is the container of a kind of loading VM in logic, can change arbitrarily the size of container by increasing or reduce VM, so not by the hardware constraints of physical machine.

The size of Vol is decided by the resource requirements such as application needs how many data transfers, how many data storages, instead of is decided by physical machine has how many virtual machines.

VM in Vol is distributed in different physical machine, as shown in Figure 4, can by assignment of traffic on different machines, and when a machine breaks down, an also only traffic affecting part, and whole business can not be affected.

Vol can increase online automatically and reduce.

Vol produces for certain is particularly applicable on certain Aggr example.

vTask

The corresponding application of vTask

VTask is made up of the vol be distributed in different Aggr

VTask defines the logic trend of the vol in different Aggr.

The relation of physical machine, virtual machine, Aggr object, Vol object and vTask object is as shown in Figure 5:

In system, each object has a global property, is used for the metamessage of data record in description object.

System is used as each VM as a dead end and is managed, all VM disposed through system have a data record, and record name, IP address, operating system version, CPU number, internal memory, hard disk size, VM are in the position of physical machine and the name of physical machine etc.

Each Aggr safeguards the allowable resource pond of a VM, is made up of doubly linked list; Also safeguard the chained list of a vol.

Each Vol safeguards a VM chained list distributed.

Each vTask safeguards an Aggr chained list distributed.

System also safeguards total physical machine chained list, the list of Aggr and the list of vTask.

System heartbeat monitors each VM belonging to Aggr, if to delay machine and VM is unreachable due to server, this VM can be stamped unavailable mark, and when VM recovers, System recover marks.

When enough hour of the allowable resource pond of VM on Aggr, add physical machine easily extensible Aggr, newly-increased VM can join the afterbody of the bi-directional list in VM pond for subsequent use.

The benefit of this design is that the level of resources utilization improves, flexible configuration, resource sharing.

Fig. 6 is the configuration before adopting, and each application is limited in physical machine, and rich resource can not be shared by the application of narrow resources or performance with performance.

The application that Fig. 7 display is disposed by this new design can share resource and the performance of VM polymerization.

When system realizes, physical resource is virtualized the resource pool for being made up of virtual machine, and virtual machine is divided into each Aggr object, and Aggr object produces the telescopic VOL variable be made up of VM.Each Aggr can have multiple Vol, Vol dynamic stretch, when Aggr space is inadequate, keeper can add VM and expand Aggr.

On Aggr example, such as generate the example of vol, by order line create vol vol1 on aggr1 vm 10, aggr1 generates vol1 example, requires system assignment 10 VM.System can physical machine different in Aggr1 select VM to generate Vol.Any one physical machine machine of delaying does not affect the application on Vol like this, adds availability.

After generating vol object instance, the example of we appointing system formation object vTask, namely according to the demand of application, specifies the vol of different levels.Such as create vTask vtask1 aggr1.vol1, aggr2.vol1, aggr4.vol1.

Here we generate the example of a vTask object, vtask1, comprise the vol1 etc. on vol1, the aggr4 on vol1, the aggr2 on aggr1.

Four. accompanying drawing explanation

Fig. 1, the Technical Architecture of typical enterprise IT software or internet, applications.

Fig. 2, the service logic framework of layering.

Fig. 3, three expressions of object on layer architecture.

Virtual machine on Fig. 4, Vol object distributes across physical machine.

Fig. 5, the relation of three objects and physical machine, virtual machine.

Fig. 6, adopts the application model before native system.

Fig. 7, adopts the application model after native system.

Fig. 8, the graph of a relation of 5 subsystems.

Five. embodiment

Cloud platform management system by 5 sub-System's composition, as shown in Figure 8:

1) resource robotization deployment system

2) virtual machine management system

3) application deployment system

4) Resourse Distribute and management system

5) resource dynamic expanding system

5.1 resource robotization deployment systems:

Resource robotization deployment system completes physical resource initialization, installation, configures and be converted into virtual resources pond.

Robotization is disposed and is also Provisioning system, is a set ofly to use Perl based on Kickstart, the system that Python and script are developed.

After switch carries out port initialization, by physical machine access network port, after machine startup, automatically guide boot by PXE, find the network segment at place, according to the requirement set, automatic selection Kickstart profile, completes automated system and installs and configuration.

While host operating system is installed, VME operating system is installed, according to the configuration template configuring virtual machine defined, configuration application service, finally enter into virtual resource standby pool, automatically the virtual machine in resource pool is divided into different polymerization clusters according to layer architecture and predefined rule base, the corresponding level of each polymerization.

5.1.2 early-stage preparations

The early-stage preparations that robotization is disposed need manually to complete, and comprising:

The configuration of BIOS and firmware, comprises the upgrading of firmware, the configuration of time, VT enable, Hyperthreadingdisable, PXE enable, Raid config etc.

Configure switch network port and the network segment, different application servers is connected to the different network segments.

Configure DNS, namely add IP and server host name.

5.1.3 operating system Auto-mounting

Can the configuration of automatic configuration system parameter after installing operating system, comprise host name, Username, Password, IP, kernel Parameters, ulimit, disable services, enable cache etc.

This system supports the agreements such as Kickstart, Systemlmager, PXE, DHCP, TFTP.

System supports Image reflection, and leaves in Systemlmager server.The first station server of application is installed and is completed by kickstart, and the configuration of other server completes robotization by Systemlmage and disposes.The time being completed robotization deployment by Systemlmager shortens 3 to 5 times than Kickstart, greatly improves installation effectiveness.

5.1.4 the Configuration and Installation of virtual machine

The Hypervisior of virtual machine selects KVM.

KVM requires that server B IOS must first enable VT.VT is the CPU operational mode of Intel expansion, can run the sensitive instructions of virtual machine at hardware-level.Owing to employing VT, KVM can simplify its code as Hypervisior greatly, and utilize the environment that the functional realiey virtual machine of Linux Kernel self requires, so KVM is smaller and more exquisite than Xen, and self runs as Kernel model, do not need other loading Kernel. as Xen

The distribution of the CPU of virtual machine:

Each virtual machine can distribute one to two physical cpu kernels

The Memory Allocation of virtual machine:

According to the demand of different business and application, virutal machine memory can be divided into 4G, 8G, 12G, 16G, 24G.

The hard disk of virtual machine distributes:

The hard disk of virtual machine is directly distributed on physical machine hard disk block subregion, can improve I/O read or write speed 3 to 5 times.

The installation of virtual machine:

Physical machine and virtual machine mixed deployment, install physical machine the postinstall stage by compression virtual machine image copy to physical machine, then decompress(ion) contract installation all virtual machines, the time of 5 to 10 times can be saved.

The configuration of virtual machine:

According to rule base partition virtual machines network, CPU, internal memory and fdisk.

5.1.5 patch upgrading (patching)

Patching system is used for the unified various operating systems for ends of the earth data center and does upgrade service.

Patching system maintenance system and the patch of software and the Repository Server of bag, can from outer off the net at various software package and patch, and the system being varying environment in Intranet provides service.

Patching system manages the version of dissimilar patch, and can be each Image patch installing or the upgrading of Image server.

Patching system selects mrepo Server to preserve patch file, and mrepo supports Yum and up2date.

5.1.6 the generation in virtual resource pond

The process of generating virtual resource pool adds record, formation object and pointer chained list etc. in table:

1) physics hangar table record is generated:

Physical machine table:

ID

Hostname

IP Addr

System

Version

Partition

VM List

2) generating virtual machines table record:

Virtual machine table:

Virtual machine:

ID，Hostname，IP，System，Version，Host ID，location，Availibility，Vol，Aggr，vTask，User，CPU，Memory，Size，Timestamp

3) by rule base, default Aggr object is generated.

Aggr object	Attribute
		Aggr0	LVS
Aggr1	Varnish
		Aggr2	Web
Aggr3	Java
		Aggr4	PHP
Aggr5	Application class 1
		Aggr6	Application class 2
Aggr7	Application class 3
		Aggr8	Memcache
Aggr9	Exploitation class

These are the predefined Aggr objects of system, and user also can customize new Aggr object and attribute thereof.

4) by rule base, resources of virtual machine is referred to different Aggr set

Virtual machine distributes and shows as shown in the figure in physical machine, supposes 10 physical machine, every platform 10 virtual machines:

Physical machine 1

Physical machine 2

Physical machine 3

Physical machine 4

Physical machine 5

Physical machine 6

Physical machine 7

Physical machine 8

Physical machine 9

Physical machine 10

P1V1

P2V1

P3V1

P4V1

P5V1

P6V1

P7V1

P8V1

P9V1

P10V1

P1V2

P2V2

P3V2

P4V2

P5V2

P6V2

P7V2

P8V2

P9V2

P10V2

P1V3

P2V3

P3V3

P4V3

P5V3

P6V3

P7V3

P8V3

P9V3

P10V3

P1V4

P2V4

P3V4

P4V4

P5V4

P6V4

P7V4

P8V4

P9V4

P10V4

..

..

..

..

..

..

..

..

..

..

P1V10

P2V10

P3V10

P4V10

P5V10

P6V10

P7V10

P8V10

P9V10

P10V10

System generates Aggr object instance, and generates Aggr storehouse table:

ID

Attribute

Physical machine list pointer

VM list pointer

Vol list pointer

vTask

At each Aggr object of generation and after being sorted out respectively by virtual machine, system generates a virtual machine standby pool for each Aggr, and represent with Two-way Cycle pointer chained list, all virtual machines are all in standby pool at the beginning,

5.2 Resourse Distribute and management system

This system manages, follows the trail of and distribute resources of virtual machine, accepts the request that following three systems are sent.

From the request of virtual machine management system

From the request of application deployment system

From the request of resource dynamic expanding system

5.2.1 the request from virtual machine management system is received:

System finds first assignable VM after freelist head in Aggr9, is returned by the ID of VM, is deleted by this VM from the standby pool circular list of Aggr9.

5.2.2 the request from application deployment system is received:

The parameter of application deployment system transmission is, application name, application architecture level, every layer of number of servers needed.

After system acceptance parameter, for this application produces a vTask example, insert vTask storehouse table, according to required framework level, for vTask distributes corresponding Aggr example and allocation order, the number of servers needed for every layer, the Aggr example of each correspondence generates the example of Vol, when generating the example of Vol, use algorithm below.

vTask ID

Aggr ID

Vol ID

VM ID chained list

Host ID chained list

System finally returns to VM name in the Vol example of Web layer and IP address thereof.

5.2.3 when the request of system acceptance from resource dynamic expanding system:

When resource dynamic expanding system monitor certain application (vTask) all very high in the load of the virtual machine (vol) of certain framework layer (Aggr), analyze through dilatation trigger mechanism, when determining dilatation, the virtual machine please be able to looked for novelty to resource allocation system, the parameter of transmission is Aggr ID and Vol ID.

Resource allocation system finds the free head in standby resources pond in the example of Aggr, find next available VM ID, judge whether the host id at this VM place exists in Vol, if existed, then look for next available VM ID, increase in VM ID to Vol, revise VM table simultaneously, indicate its vol, aggr and vTask, return vol to resource dynamic expanding system.

5.3 resource dynamic expanding systems

The load of resource dynamic expanding system monitoring and measuring application program and virtual machine, when determining dilatation, vol ID is sent to Resourse Distribute and management system, the latter returns after increasing VM to Vol.

5.3.1 monitoring

By heartbeat monitoring system accessibility, by monitoring the load of virtual machine at monitoring host computer and the Agent that disposes in physical machine and virtual machine, comprise the data such as CPU, internal memory, process number, I/O, swap, ping, serve port linking number, log, these information of periodic collection.

5.3.2 trigger mechanism

To each vTask, from the Aggr of the second level, the monitor message of all virtual machines in systematic collection vol, next step will determine whether vol dilatation or capacity reducing, we judge although can do rule of thumb to some rules of system, but the change of various parameter is very complicated in actual motion, if rule is cured in program, then the change of actual conditions cannot be adapted to.And when input parameter is a lot, even if a lot of rules also may not necessarily the various possibility of limit.

And artificial neural network may address this problem well, artificial neural network belongs to artificial intelligence field, can simulate nonlinear computation well.Here adopt the artificial neural network of sandwich construction, the input and output relation of network can be expressed as Y=XxW simply, and wherein X is one dimension input variable, and Y is one dimension output variable, and W is two-dimentional weight matrix.

Regulate matrix full weight to adapt to practical application to enable network, this process is called study or the training of artificial neural network.Here BP (i.e. Back-Propagation) learning algorithm is selected to train, the study of BP network is a kind of study having supervision, when network training, training sample (data) is selected from actual application environment, training data is paired input and output, input amendment data are added to network input, compared with actual to corresponding output sample and network output, obtain error signal, with the adjustment of this control weight strength of joint simultaneously.

Train time, once all training samples to all reading in internal memory, then circulate and use often pair of training sample training network successively, until error is less than defined threshold.Then we to obtain the strength of joint of each layer of network, threshold value and study constant equivalent.

Here is specific algorithm:

Here in order to be judged accurately, a large amount of training samples must be received, namely collect performance and the load parameter of a large amount of VM, and need the result manually providing dilatation or capacity reducing, so the time of collecting sample can be long.But once collect enough sample datas all, due to the high-performance of modern computer, the time of training is not long.

When actual motion, the Monitoring Data of each VM of collected Vol is input to the input end of neural network by system, and network exports <-1,0, > 1 three numerical value.<-1 just represents capacity reducing, and 0 representative maintains, and > 1 represents dilatation.

If there is the output of the VM of 80% to be > 1 in Vol, just determine Vol dilatation.

5.3.3 dynamic capacity-expanding

When system determines dilatation or capacity reducing by the degree of sentencing of artificial neural network, system increases to resource allocation system application or reduces VM, and transmit Aggr and Vol parameter to resource allocation system, resource allocation system returns vol parameter, wherein has VM that is newly-increased or that reduce.

If newly-increased VM, the VM configuration that system copies keeps in advance copies in new VM.

System finds all VM of the vol in Aggr (i-1), watch service or the application of each VM, if standards service such as LVS, Apache, PHP, Java, the configuration file of this service is found out in service configuration table, the IP of new VM name increased, the configuration file that the order of specifying according to the corresponding table of service makes service read again to revise.

Such as to LVS, run ipvsadm-a-t $ VIP:$ porr-r $ RS-g

Service

Configuration file

Startup command

The corresponding table of service

If application program:

1) in application allocation list, find corresponding configuration template, increase VM and IP record

2) Agent in VM can watch configuration template in the mode of poll

3) once find that file changes (version or timestamp), Agent again reads configuration template and generates new application profiles,

4) and by the signal transmission mode of Inter-Process Communication notify to apply and read application configuration file again, as kill-hup Pid

If application program can not read configuration file again, then the shared drive mode of Inter-Process Communication must be used to notify application.

5.3.4 dynamic capacity reducing

If load reduces, system judges capacity reducing by artificial neural network, then system finds all VM of the vol of Aggr (i-1) in the iTask of vol place, the configuration file of each VM is revised in the mode of similar dilatation, last VM and IP is deleted from configuration file, and reads configuration file again with the mode notification service of similar dilatation or application.

The ID of Aggr and vol is passed to resource allocation system by system, and the latter revises the VM list of vol, is taken away by last VM from vol, adds the afterbody of the doubly linked list of standby pool freelist head to, and returns the pointer of vol.

5.3.5 fault and recovery

Fault category has:

Server hardware fault, as CPU, internal memory, motherboard, hard disk, network interface card damage

Server system failure, as operating system collapse, virtual HyperVisor collapse etc.

The system failure of VM, as VME operating system collapse, service stopping

Software fault on VM, as software bug, under attack

Unavailable classification:

On hardware server, all VM are unavailable

Part VM is unavailable

A VM is unavailable

To delay machine if system obtains physical machine by supervisory system detecting, show the VM finding it to have by the storehouse of physical machine, the available flag position of the relevant VM in corresponding VM table and vTask, Aggr, Vol being shown is set to unavailable, and setup times stabs.

If unavailable time was more than one hour, system finds all VM on unavailable VM place vTask-> Aggr (i-1)-> vol, according to service configuration table, find configuration file, save as previous release, fault VM and IP deletes by amendment configuration file, and configuration file is read in the order notification service of specifying according to service configuration table or application again.

If unavailable time exceedes as 24 hours, physical machine record is deleted by system, the VM pointer deleted VM record simultaneously and be correlated with in chained list.

If unavailable time surpasses after an hour but recovered in 24 hours, system judges whether application program and data lose, if lost, then same by the deletion of VM record; If do not lost, all VM on VM place vTask-> Aggr (the i-1)-> vol that system finds recovery, according to service configuration table, find the previous release of configuration file, cover current version, the order notification service of specifying according to service configuration table or application program read configuration file again.

5.4 application deployment system:

The external application service that this system both can have been served, also can to inner application program service, as search, filtration, data mining, data analysis etc.

This system is by web services interface, foreground, collect application name, application architecture level, every layer of number of servers needed that user submits to, information is given to Resourse Distribute and management system, Resourse Distribute and management system generate the example of the Vol of vTask and every layer of Aggr, and return the ID. of vTask, Aggr and Vol

Application deployment system finds out web layer Aggr, returns the web URL of VM in its vol to user.

User uploads deployment and configuring application program by URL, configuration database.

After user completes, system generates configuration template reflection, and deposits in vol.image.

5.5 virtual machine management systems:

Virtual machine management system is for programmer's management and distribute exploitation virtual machine.

Development of virtual machine manages distribution in Aggr9.

Development of virtual machine and programmer are one to one.

After system receives the request of user's submission by web interface, generator program person's record.

System

Version

Architect

Programming

Application

Department

User

DB

Duration

VM

The workflow portion of system and office automation and workform management system interface, produce a workflow, arrive first software division supervisor, arrive network system portion supervisor again, supervisor ratifies and after putting on record, produce a shop order, arrive network system portion workform management system, network system department employee selects this shop order.

System manager's login management web interface, selects Next virtual machine, distributes to programmer from the virtual resource pond for subsequent use Aggr9 (development environment).

The next VM of the standby resources pond doubly linked list Freelist head of Aggr9 deletes by system, and is filled into by this VM ID in the VM item of programmer's record.

When application and development completes or programmer's leaving office, the recyclable developing engine of system manager, system-kill programmer record, adds the afterbody of Aggr9 standby resources pond doubly linked list to by VM ID.

Claims (3)

1. the multi-application system with layer architecture managed based on 3 kinds of object logics shares a dynamic self extending management method for cloud computing resource pool, it is characterized in that comprising:

Described layer architecture and layering deployment system framework, refer to when application system is deployed on cloud computing platform, according to the Service Design of application system, system platform needs the system architecture of layering, comprise the framework level of LVS load balancing layer, global buffer layer, Web Cluster layer, application layer, database caches layer, database layer and other application system definition, can need to use 1 in each layer to multiple with server or virtual machine be unit calculating and storage resources;

Described multi-application system is shared, refer to that each application system has layering deployment system framework as above, but the shared level of each application system is incomplete same, multiple application system with layer architecture can share framework layer, in the framework layer shared, shared the physical equipment resource on same level by Intel Virtualization Technology;

The first of described three kinds of object logics, refer to that in cloud computing resource pool, all physical servers being disposed the resource required for running according to application system carries out framework layering, after layering, all physical machine configurations of every one deck are identical, all for each level physical machine are arranged to multiple virtual machine by Intel Virtualization Technology, all resources of virtual machine of every one deck represent with an object logic, be called as Aggr, Aggr represents this layer of upper all set distributing to the virtual machine of application system, the assignable virtual machine list of in framework i-th layer that what Aggr (i) comprised is,

The recording mode of virtual machine in Aggr (i) adopts the data structure of bi-directional list, head node is a pointer, virtual machine can be distributed below according to first of i-th layer of first physical machine, first of second physical machine can be distributed virtual machine, first of last physical machine can be distributed virtual machine, second of first physical machine can be distributed virtual machine, second of second physical machine can be distributed virtual machine, second of last physical machine can be distributed virtual machine, until the order of last distributed virtual machine of i-th layer of last physical machine records and mutually points to, whenever distributing to application system virtual machine, the distributed virtual machine of Aggr just reduces one, head node can move backward and move a step,

The second of described three kinds of object logics, refer on the system architecture layer at each Aggr object place, for the resources of virtual machine set distributed when different application systems is disposed, the virtual machine set that Aggr distributes to an application is expressed as Vol with an object logic, Vol can be regarded as an application system loads the container of virtual machine on each deployment framework level, the size of container can be changed arbitrarily by increasing or reduce virtual machine quantity; The size of Vol depends on application system all the time to the demand of resource; Virtual machine in Vol is distributed in different physical machine, can be assigned on different machines by flow and load, when a machine breaks down, can not affect whole business; Vol can increase online automatically and reduce, as long as Aggr has enough spaces;

The third of described three kinds of object logics, refer to cloud computing resource pool share by multiple application system time, by abstract for the set of the Vol of each application system on each used system architecture layer and Aggr layer be an object logic to be expressed as vTask, simultaneously vTask object defines the service logic trend of the Vol between different Aggr layer in inside.

2. the method for claim 1, possesses virtual computing distribution of resource formula and disposes and smart allocation ability, and possess Fault Tolerance, it is characterized in that following steps:

The first step, during system initialization, the physical machine chained list that system maintenance is total, and the example generating Aggr object according to framework layering;

Second step, after the physical machine of Aggr place level is virtual, each virtual machine VM is represented, each VM has a data record, record name, IP address, operating system version, CPU number, internal memory, hard disk size, the position of VM in physical machine and the name of physical machine, by these unappropriated VM with doubly linked list be recorded in Aggr across the distributed form of machine;

3rd step, management system monitoring belongs to each VM of Aggr, if cause VM unreachable due to the server machine of delaying, this VM can be stamped unavailable mark, and when VM recovers, System recover marks; When enough hour of the quantity of VM on Aggr, add physical machine easily extensible Aggr, newly-increased VM can join the afterbody of the bi-directional list in VM pond in Aggr for subsequent use;

4th step, when application system needs in cloud computing resource pool deploy, the example of management system formation object vTask, namely disposes the demand of framework according to application system, specifies Aggr object and the Vol object of different levels;

5th step, management system is be applied in the example in vTask, each Aggr example generating Vol, and between different Aggr, specify the logic of Vol to move towards;

Virtual machine in described Aggr is according to the journal across physical machine, therefore naturally defining the resources of virtual machine that each application system is assigned to when each level deploy can be distributed in different physical machine, both can carry out load sharing with this distributed deployment, also can improve the availability of system.

3. the method for claim 1, when resource needs expansion, possesses intelligent decision, automatically triggers, the ability of automatic learning and dynamic expansion, it is characterized in that following steps:

The first step, by monitoring application and system load index, input artificial neural network two-dimensional matrix, utilizes backpropagation BP algorithm, to the training of neural network matrix, allows system automatic decision and autonomous learning increase, the trigger condition of constant or minimizing virtual machine;

Second step, if increase virtual machine, only needs in Aggr, find next assignable virtual machine pointer be given to Vol and delete in Aggr; If minimizing virtual machine, in Vol, find last virtual machine added and delete from Vol, adding in Aggr afterbody;

3rd step, management system can check the service of all VM in the Vol of Aggr (i-1) layer, amendment service profiles, increase or reduce name and the IP address of VM, and make to serve the configuration file read again and revised according to the order that the corresponding table of service is specified, thus realize online dynamically change.