CN104683480A - Distribution type calculation method based on applications - Google Patents
Distribution type calculation method based on applications Download PDFInfo
- Publication number
- CN104683480A CN104683480A CN201510122458.9A CN201510122458A CN104683480A CN 104683480 A CN104683480 A CN 104683480A CN 201510122458 A CN201510122458 A CN 201510122458A CN 104683480 A CN104683480 A CN 104683480A
- Authority
- CN
- China
- Prior art keywords
- server
- cluster
- instance
- resource
- branch
- 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
Links
Classifications
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
- H04L67/00—Network arrangements or protocols for supporting network services or applications
- H04L67/01—Protocols
- H04L67/10—Protocols in which an application is distributed across nodes in the network
- H04L67/1001—Protocols in which an application is distributed across nodes in the network for accessing one among a plurality of replicated servers
Abstract
The invention provides a distribution type calculation method based on applications. The distribution type calculation method comprises the following steps: maintaining a server instance of each cluster and the state of an application in a distributed cache manner; when the service instance needs to migrate, shutting off original progress at a present server, starting new progress at a target server, and modifying a forwarding list loaded with a distribution server; enabling each server instance to operate in the server in a system progress operation manner; receiving a periodic state report of a branch server node through a main server node, and performing resource integration; monitoring and reporting property data of the server instances in operation through a plurality of branch server nodes, and adjusting through the forwarding list loaded with the distribution server. According to the distribution type calculation method, sharing of multiple lease accepting resources in a cluster distribution type environment is achieved, server resource waste or overloading can be prevented, and the utilization rate is improved.
Description
Technical field
The present invention relates to Distributed Calculation, particularly a kind of distributed computing method based on application.
Background technology
Current mainstream server all supports clustering functionality, and traditional server cluster runs an example on every platform physical server.Because this scheme is to tackle peak load, not high in the utilance of many times server.PaaS provides computational resource with the form of platform service for applying, shield complexity and the isomerism of bottom cloud facility, more upper layer application provides flexible resource to provide, charter the cloud computing key characteristic supports such as resource-sharing more, final realization ensures application performance, the object of efficiency utilization platform resource.The IT infrastructure reducing tenant drops into, and improves the resource utilization of data center.Cloud platform need to guarantee tenant apply between performance isolation, and carry out suitable Server Consolidation according to Current resource utilization state.But the isolation method of current cluster is comparatively large on the impact of cluster instance migration, the degree of resource-sharing is not high, easily produces server overload, affects application performance.
Summary of the invention
For solving the problem existing for above-mentioned prior art, the present invention proposes a kind of distributed computing method based on application, comprising:
In logical layer, system comprises multiple server cluster, the corresponding application of each cluster, and each server cluster comprises a load distribution server and one or more stateless server instance, and the state of application adopts distributed caching to safeguard, when server instance needs migration, only close original process at current server, start new process at destination server, revise the forwarding list of cluster load distribution server simultaneously, system comprises multiple servers within the physical layer, and as the running environment of cluster, each server instance all runs in the server in the mode of operating system process, run one or more server instance coming from multiple cluster on the same server, these examples share the service and resource that same operating system provides, and all application clusters share identical server environment in global scope, server is divided into primary server joint and branch server node within the physical layer, and primary server joint for receiving the periodic status report of branch server node, and is implement resource integration, multiple branch server node is for monitoring and reporting the performance data of server instance run, receive simultaneously and perform the control command of primary server joint, namely the resource adjustment operation of cluster is all calculated by primary server joint, and control command is mail to the execution of branch server node, forwarding list change finally by load distribution server adjusts, the server instance of a same cluster can only be there is in same station server, the resource service condition of branch server monitoring nodes server and cluster example, and the information obtained regularly is sent to primary server joint
The present invention compared to existing technology, has the following advantages:
The present invention is used for chartering resource-sharing under cluster distributed environment more, prevents server resource from wasting and overload, increases operation rate.
Accompanying drawing explanation
Fig. 1 is the flow chart of the distributed computing method based on application according to the embodiment of the present invention.
Embodiment
Detailed description to one or more embodiment of the present invention is hereafter provided together with the accompanying drawing of the diagram principle of the invention.Describe the present invention in conjunction with such embodiment, but the invention is not restricted to any embodiment.Scope of the present invention is only defined by the claims, and the present invention contain many substitute, amendment and equivalent.Set forth many details in the following description to provide thorough understanding of the present invention.These details are provided for exemplary purposes, and also can realize the present invention according to claims without some in these details or all details.
The present invention realizes the server cluster system towards PaaS.And from logical construction and physical structure two layers, cluster architecture is introduced.An aspect of of the present present invention provides a kind of distributed computing method based on application.Fig. 1 is the distributed computing method flow chart based on application according to the embodiment of the present invention.
At logical layer, system comprises multiple server cluster, the corresponding application of each cluster, each server cluster comprises a load distribution server and one or more stateless server instance, and apply correlation behavior and usually adopt distributed caching to safeguard, server instance fault can not affect application availability.When server instance needs migration, only need to start new process at destination server, close original process at current server, revise the forwarding list of cluster load distribution server simultaneously.Compare virtual machine (vm) migration, such moving method is very little to application performance impact.
In physical layer, system comprises multiple servers, and as the running environment of cluster, each server instance operates in these servers in the mode of operating system process; Can run one or more server instance coming from multiple cluster on the same server, these examples can share the service and resource that same operating system provides, and in global scope, all application clusters share identical server environment.
Within the physical layer, two class server nodes are comprised: primary server joint and branch server node.Primary server joint for receiving the periodic status report of branch server node, and is implement resource integration; Multiple branch server node, for monitoring and reporting the various performance datas of server instance run, receives simultaneously and performs the control command of primary server joint.Therefore, the resource adjustment operation of cluster is all calculated by primary server joint, and control command is mail to the execution of branch server node, and the forwarding list change finally by load distribution server reaches the effect of adjustment.In order to diversify risks and avoid the wasting of resources, the server instance of the same cluster of application deployment constraint requirements can only exist one in same station server.
The resource consolidation of system is periodic behavior, and the self adaptation realizing resource provides.The resource service condition of branch server monitoring nodes server and cluster example, and the information obtained regularly is sent to primary server joint.Primary server joint uses following determination methods, searches and is in Idle state and is in the server of overload state, and considers resource and the load Distribution situation of the overall situation, formulates the resource comprising a series of adjustment behavior and provides strategy; Resource reclaim manager is distributed in adjustment behavior with resource management assembly and resource provides manager to perform; Finally rely on the load distribution server that each cluster provides, load is redistributed.
The target that server resource is integrated reduces the usage quantity of server in host environment, reduces energy resource consumption while being to ensure that in group system, each application can normally run, the operation cost of reduction data center.It realizes mainly through the integration of idle server and the division of Overloaded Servers.
Server Consolidation needs server set idle in first seeking system, for each cluster example above selects a suitable destination server, under the prerequisite ensureing system cloud gray model constraint, by these cluster instance migration in destination server, originally idle server can be closed or enter the power save modes such as standby.If the server environment of system uses Intel Virtualization Technology to build, then direct these virtual machines to be regained.
Server divides for the treatment of the server being in overload, therefrom select the part cluster example needing to move away, and these cluster examples are re-deployed in other servers of system, thus eliminate the overload of these servers, reduce the resource contention between cluster example in this server, realize performance isolation with this.
Hereafter with S
1, S
2... S
nrepresent n physical server in group system, ssize (S
i) representative server S
ithe server instance quantity of upper operation, S
ijrepresentative is at S
ia jth cluster example of upper operation.KS
irepresent the Current resource use amount of i-th server, KS
ijrepresent S
ithe Current resource use amount of a jth cluster example, TKS
irepresent the Current resource total amount of i-th server.Resource herein can refer to the weight summation of one or several data of the indexs such as CPU, internal memory, network read/write, disk read/write.
Server and cluster example state are simply judged, then selects suitable resource integrated method according to judged result.Symbol full and idle is used to describe judged result herein:
If KS
i/ TKS
i>0.8, namely resource utilization is more than 80%, then full (S
i)=true, there is resource overload in server, needs to carry out server division.If KS
i/ TKS
i<0.3, the i.e. low utilization of resources 30%, then idle (S
i)=true, server free, possesses the condition of Server Consolidation.
Server Consolidation to as if all servers being in idle condition, they are integrated into less server, thus reduce the quantity of workspace server, cluster example in n station server is redistributed in m station server, reduce the size of m value as far as possible, and ensure that this m station server there will not be overload.
In Server Consolidation process, by some cluster examples wherein from a server migration to another one server.Because clustered node is stateless, so first identical cluster example can be started at destination server, after success, then the cluster example in current server is closed.The load distribution server of cluster can be modified to its forwarding list, by request forward in new clustered node, realizes the migration of clustered node.
When carrying out Server Consolidation, considering due to the cost that cluster instance migration is brought, reducing the movement of cluster example.Avoiding same cluster instance migration repeatedly, needing the result to integrating to predict simultaneously, there is overload in the destination server after preventing from integrating.Performance parameter, as CPU, network, response time etc.
First integration process is in the server set of idle condition in seeking system, and carries out descending.By the cluster instance migration on the server of position rearward in server on the front, minimize cluster instance migration number of times, reduce system change, keep the stable of each application cluster and system.Add up the total resources of the current use of all cluster examples, minimum number of servers serversize is searched in server list, ensure that the total amount of often kind of resource exceedes 1.5 times that all cluster examples take total resources, then can all cluster examples all be moved in this serversize server as far as possible.Migration S
iall cluster examples in (serversize<i≤n) are in front serversize station server.The stock number expected calculated destination server migration before migration cluster example after, and the server that selective value is minimum, go over cluster instance migration, and occur overload after preventing from moving.
Server divide to as if there is the server that transships, object is divided in other server by part cluster example to run, eliminate the overload of this server, improves overall operational efficiency.
Partition process can be divided into following step:
(1) all server set full_servers needing to divide are searched.
(2) for any S
i∈ full_servers, searches the cluster example collection cluster_nodes [S needing to move away
i], make these cluster instance migration go out rear server and can be in busy and non-overloaded level.
(3) to each clustered node S
ij∈ cluster_nodes [S
i], by cluster example S
ijbe placed in other servers of platform.
Reduce branch server node S as far as possible
ithe cluster example collection cluster_nodes [S divided away
i], to reduce the cost of cluster instance migration, therefore require cluster_nodes [S
i] in each cluster example there is higher resource use amount, bear more load.But for the clustered node that load is higher, owing to needing to shift a large amount of user loads, make migration cost higher, in transition process, certain impact can be produced on the service quality of application, therefore require again the clustered node that migration load is lower.Integrate, pay the utmost attention to the quantity of migration clustered node, secondly consider the low load node of migration.Suppose N=ssize (S
i), then complexity is O (N total time
2).
In sum, the present invention is used for chartering resource-sharing under cluster distributed environment more, prevents server resource from wasting and overload, increases operation rate.
Obviously, it should be appreciated by those skilled in the art, above-mentioned of the present invention each module or each step can realize with general computing system, they can concentrate on single computing system, or be distributed on network that multiple computing system forms, alternatively, they can realize with the executable program code of computing system, thus, they can be stored and be performed by computing system within the storage system.Like this, the present invention is not restricted to any specific hardware and software combination.
Should be understood that, above-mentioned embodiment of the present invention only for exemplary illustration or explain principle of the present invention, and is not construed as limiting the invention.Therefore, any amendment made when without departing from the spirit and scope of the present invention, equivalent replacement, improvement etc., all should be included within protection scope of the present invention.In addition, claims of the present invention be intended to contain fall into claims scope and border or this scope and border equivalents in whole change and modification.
Claims (1)
1., based on a distributed computing method for application, for implemening resource integration in the server cluster system of PaaS, it is characterized in that, comprising:
In logical layer, system comprises multiple server cluster, the corresponding application of each cluster, and each server cluster comprises a load distribution server and one or more stateless server instance, and the state of application adopts distributed caching to safeguard, when server instance needs migration, only close original process at current server, start new process at destination server, revise the forwarding list of cluster load distribution server simultaneously, system comprises multiple servers within the physical layer, and as the running environment of cluster, each server instance all runs in the server in the mode of operating system process, run one or more server instance coming from multiple cluster on the same server, these examples share the service and resource that same operating system provides, and all application clusters share identical server environment in global scope, server is divided into primary server joint and branch server node within the physical layer, and primary server joint for receiving the periodic status report of branch server node, and is implement resource integration, multiple branch server node is for monitoring and reporting the performance data of server instance run, receive simultaneously and perform the control command of primary server joint, namely the resource adjustment operation of cluster is all calculated by primary server joint, and control command is mail to the execution of branch server node, forwarding list change finally by load distribution server adjusts, the server instance of a same cluster can only be there is in same station server, the resource service condition of branch server monitoring nodes server and cluster example, and the information obtained regularly is sent to primary server joint.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201510122458.9A CN104683480A (en) | 2015-03-19 | 2015-03-19 | Distribution type calculation method based on applications |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201510122458.9A CN104683480A (en) | 2015-03-19 | 2015-03-19 | Distribution type calculation method based on applications |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN104683480A true CN104683480A (en) | 2015-06-03 |
Family
ID=53318046
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201510122458.9A Pending CN104683480A (en) | 2015-03-19 | 2015-03-19 | Distribution type calculation method based on applications |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN104683480A (en) |
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106484379A (en) * | 2015-08-28 | 2017-03-08 | 华为技术有限公司 | A kind of processing method and processing device of application |
| CN106844035A (en) * | 2017-02-09 | 2017-06-13 | 腾讯科技（深圳）有限公司 | A kind of method and device realized the release of Cloud Server resource or recover |
| CN107085577A (en) * | 2016-02-15 | 2017-08-22 | 华为技术有限公司 | A kind of method, calculate node and coordinator node for loading data page |
| CN110083504A (en) * | 2019-03-29 | 2019-08-02 | 北京奇安信科技有限公司 | The running state monitoring method and device of distributed task scheduling |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101505243A (en) * | 2009-03-10 | 2009-08-12 | 中国科学院软件研究所 | Performance exception detecting method for Web application |
| CN102014169A (en) * | 2010-12-22 | 2011-04-13 | 北京中电普华信息技术有限公司 | Distributed service system as well as distributed service system task execution method and device |
| CN102130938A (en) * | 2010-12-03 | 2011-07-20 | 中国科学院软件研究所 | Resource supply method oriented to Web application host platform |
| CN103353852A (en) * | 2013-06-07 | 2013-10-16 | 曲阜师范大学 | Method for constructing IaaS of virtualized WebService |
-
2015
- 2015-03-19 CN CN201510122458.9A patent/CN104683480A/en active Pending
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101505243A (en) * | 2009-03-10 | 2009-08-12 | 中国科学院软件研究所 | Performance exception detecting method for Web application |
| CN102130938A (en) * | 2010-12-03 | 2011-07-20 | 中国科学院软件研究所 | Resource supply method oriented to Web application host platform |
| CN102014169A (en) * | 2010-12-22 | 2011-04-13 | 北京中电普华信息技术有限公司 | Distributed service system as well as distributed service system task execution method and device |
| CN103353852A (en) * | 2013-06-07 | 2013-10-16 | 曲阜师范大学 | Method for constructing IaaS of virtualized WebService |
Non-Patent Citations (1)
| Title |
|---|
| 赵鑫等: ""共享式Web应用服务器集群的资源整合方法研究"", 《计算机科学与探索》 * |
Cited By (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106484379A (en) * | 2015-08-28 | 2017-03-08 | 华为技术有限公司 | A kind of processing method and processing device of application |
| CN106484379B (en) * | 2015-08-28 | 2019-11-29 | 华为技术有限公司 | A kind of processing method and processing device of application |
| CN107085577A (en) * | 2016-02-15 | 2017-08-22 | 华为技术有限公司 | A kind of method, calculate node and coordinator node for loading data page |
| CN107085577B (en) * | 2016-02-15 | 2021-01-05 | 华为技术有限公司 | Data page loading method, computing node and coordination node |
| CN106844035A (en) * | 2017-02-09 | 2017-06-13 | 腾讯科技（深圳）有限公司 | A kind of method and device realized the release of Cloud Server resource or recover |
| CN110083504A (en) * | 2019-03-29 | 2019-08-02 | 北京奇安信科技有限公司 | The running state monitoring method and device of distributed task scheduling |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN104679594B (en) | A kind of middleware distributed computing method | |
| CN108182105B (en) | Local dynamic migration method and control system based on Docker container technology | |
| Tziritas et al. | Application-aware workload consolidation to minimize both energy consumption and network load in cloud environments | |
| US8914513B2 (en) | Hierarchical defragmentation of resources in data centers | |
| Gao et al. | An energy-aware ant colony algorithm for network-aware virtual machine placement in cloud computing | |
| CN105245617A (en) | Container-based server resource supply method | |
| Xu et al. | Migration cost and energy-aware virtual machine consolidation under cloud environments considering remaining runtime | |
| KR101578177B1 (en) | Method and system for migration based on resource utilization rate in cloud computing | |
| CN110221920B (en) | Deployment method, device, storage medium and system | |
| CN104050042A (en) | Resource allocation method and resource allocation device for ETL (Extraction-Transformation-Loading) jobs | |
| Patni et al. | Load balancing strategies for grid computing | |
| US9274841B2 (en) | System for energy saving in company data centers | |
| Singh et al. | Survey on various load balancing techniques in cloud computing | |
| CN103023936B (en) | Multi-hierarchy network system and task executing method based on same | |
| Alghamdi et al. | Profit-based file replication in data intensive cloud data centers | |
| Bourhim et al. | Inter-container communication aware container placement in fog computing | |
| Fu et al. | Network traffic based virtual machine migration in cloud computing environment | |
| CN104683480A (en) | Distribution type calculation method based on applications | |
| Deiab et al. | Energy efficiency in cloud computing | |
| Chaudhary et al. | An analysis of the load scheduling algorithms in the cloud computing environment: A survey | |
| Jaiswal et al. | An approach towards the dynamic load management techniques in cloud computing environment | |
| Guo | Ant colony optimization computing resource allocation algorithm based on cloud computing environment | |
| Vinothini et al. | Meta-heuristic firefly approach to multi-servers load balancing with independent and dependent server availability consideration | |
| Butt et al. | Optimization of response and processing time for smart societies using particle swarm optimization and levy walk | |
| Swarnakar et al. | A novel improved hybrid model for load balancing in cloud environment |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20150603 |
|
| RJ01 | Rejection of invention patent application after publication |