CN103763130B

CN103763130B - Management method, the device and system of large-scale cluster

Info

Publication number: CN103763130B
Application number: CN201310752189.5A
Authority: CN
Inventors: 王黎; 吴晓明
Original assignee: Huawei Digital Technologies Suzhou Co Ltd
Current assignee: Huawei Digital Technologies Suzhou Co Ltd
Priority date: 2013-12-31
Filing date: 2013-12-31
Publication date: 2018-06-19
Anticipated expiration: 2033-12-31
Also published as: CN103763130A; WO2015101089A1

Abstract

The embodiment of the present invention provides a kind of management method of large-scale cluster, device and system, can carry out performance management and scheduling of resource to user according to the grade of service, improve user experience.This method includes：At least one management object is determined in the corresponding management object of first service grade of multiple grades of service, wherein management object is the resource unit in large-scale cluster；Determine the target capabilities of at least one management object；Obtain the actual performance of at least one management object；Performance management is carried out according to target capabilities and actual performance management object corresponding to first service grade.The embodiment of the present invention in the corresponding management object of the first service grade in large-scale cluster by determining at least one management object, and carry out performance management according to target capabilities and actual performance all management objects corresponding to the first service grade of at least one management object, performance so as to ensure most even whole users reaches target capabilities, improves user experience.

Description

Management method, the device and system of large-scale cluster

Technical field

The present invention relates to field of cloud calculation, and the management method more particularly, to large-scale cluster, device and it is System.

Background technology

With the further development and the requirement of mass data computing capability of computer network, various mass computing abilities Computer hardware continuously emerges.In addition, global information system WWW is also very popular.These software and hardware technologies or equipment go out It is existing, to propose a kind of novel referred to as " cloud computing（Cloud Computing）" computation model provide possibility.

The cloud computing of narrow sense refers to information technology（Information Technology, referred to as " IT "）The friendship of infrastructure Pay and use pattern, refer to by network with on-demand, easy extension way obtain needed for resource；The network for providing resource is referred to as " cloud（Cloud）”.Resource in " cloud " is appeared to be in user can be with infinite expanding, and can obtain at any time, expands at any time Exhibition uses, and pay-per-use on demand.

The cloud computing of broad sense refers to delivery and the use pattern of service, refers to and is obtained by network with on-demand, easy extension way Required service.This service can to IT, software, internet are related or other are serviced, the network for providing service is claimed For " cloud（Cloud）”." cloud " is that some can be with self and the virtual computing resource of management, usually some large servers Cluster, including calculation server, storage server, broadband resource etc..Cloud computing to largely use network connection computing resource into Row unified management and scheduling, form a computing resource pond, to provide a user on-demand service.

Since cloud computing has the spies such as ultra-large, virtualization, high reliability, versatility, high scalability, on-demand service Property, cloud computing is more and more widely paid close attention to.

In cloud computing application, cloud computation data center conformity calculation resource, storage resource and Internet resources, using virtual The technologies such as change and pass through network user is supplied to use.The form of application can include virtual machine（Virtual Machine, letter Referred to as " VM "）, storage volume etc..Virtualization technology is by generating the applications such as large-scale virtual machine and large-scale storage volume, structure Into extensive large-scale cluster.Performance management how is carried out to extensive large-scale cluster and experience ensures increasingly to need The problem of paying close attention to.

The management of existing extensive large-scale cluster is usually with server（Server）, resource pool（Pool）Even cluster （Cluster）For unit, a small amount of money performance management even if as unit of user corresponding to only for a small number of VIP users Source, in this way, the performance management of most of user can not be guaranteed that user experience is poor.

Invention content

The embodiment of the present invention provides a kind of management method of large-scale cluster, device and system, can be according to the grade of service Performance management and scheduling of resource are carried out to user, improve user experience.

In a first aspect, a kind of management method of large-scale cluster is provided, including：In the first service of multiple grades of service At least one management object is determined in the corresponding management object of grade, wherein the management object is in the large-scale cluster Resource unit；Determine the target capabilities of at least one management object；Obtain the practical property of at least one management object Energy；Performance pipe is carried out according to the target capabilities and the actual performance management object corresponding to the first service grade Reason.

With reference to first aspect, in the first realization method of first aspect, first clothes in multiple grades of service Before at least one management object is determined in the corresponding management object of grade of being engaged in, further include：It is institute according to service-level agreement SLA The management object stated in large-scale cluster determines the multiple grade of service.

With reference to first aspect and its above-mentioned realization method, it is described according to SLA in second of realization method of first aspect After determining multiple grades of service for the management object in the large-scale cluster, further include：Determine the multiple grade of service The target capabilities of middle first service grade；The target capabilities for determining at least one management object, including：By described The target capabilities of one grade of service are determined as the target capabilities of at least one management object.

With reference to first aspect and its above-mentioned realization method, in the third realization method of first aspect, the determining institute The target capabilities of at least one management object are stated including at least one of following：According to determining scheduled performance strategy extremely Few one manages the corresponding target capabilities of object；Or the Objective of at least one management object is manually set Energy.

With reference to first aspect and its above-mentioned realization method, in the 4th kind of realization method of first aspect, the Objective The type of energy includes at least one of response delay, read-write number IOPS per second, message transmission rate, CPU usage.

With reference to first aspect and its above-mentioned realization method, in the 5th kind of realization method of first aspect, the acquisition institute The actual performance of at least one management object is stated, including：Monitor periodically or routinely at least one management object Actual performance.

With reference to first aspect and its above-mentioned realization method, it is described according to institute in the 6th kind of realization method of first aspect It states target capabilities and the actual performance management object corresponding to the first service grade and carries out performance management, including：Really Whether the actual performance got surely meets the target capabilities；The target capabilities are unsatisfactory in the actual performance When, manage corresponding to the first service grade removes the first service grade in object and/or the multiple grade of service The corresponding management object of other grades of service carry out the performance management so that the actual performance of the first service grade Meet the target capabilities.

With reference to first aspect and its above-mentioned realization method, in the 7th kind of realization method of first aspect, the performance pipe Reason includes at least one of following：Business migration；Business limits；Flow control；Scheduling of resource；Send out alarm.

With reference to first aspect and its above-mentioned realization method, in the 8th kind of realization method of first aspect, in the reality When performance meets the target capabilities, the corresponding management object of first service grade in multiple grades of service is repeated In determine it is at least one management object the step of or repeat it is described obtain it is described it is at least one management object practical property The step of energy.

With reference to first aspect and its above-mentioned realization method, it is described multiple in the 9th kind of realization method of first aspect At least one management object is determined in the corresponding management object of first service grade of the grade of service, including：In the described first clothes At least one management object for meeting predetermined condition is determined in the corresponding management object of business grade, wherein the predetermined condition includes At least one of settling time, location information, loading condition and historical record；Or according to pre-defined algorithm in the described first clothes At least one management object is determined in the corresponding management object of grade of being engaged in, wherein the pre-defined algorithm is including randomly selecting, sequence At least one of selection, time choice of dynamical.

With reference to first aspect and its above-mentioned realization method, in the tenth kind of realization method of first aspect, the management pair As including virtual machine VM, storage volume, virtual switch vSwitch, virtual local LAN vLAN, input and output I/O ports, handing over It changes planes, at least one of network bandwidth and server.

Second aspect provides a kind of managing device of large-scale cluster, including：Determination unit, in multiple services At least one management object is determined in the corresponding management object of first service grade of grade, wherein the management object is described Resource unit in large-scale cluster；The determination unit is additionally operable to determine the target capabilities of at least one management object； Acquiring unit, for obtaining the actual performance of at least one management object；Capability management unit, for according to the target Performance and actual performance management object corresponding to the first service grade carry out performance management.

With reference to second aspect, in the first realization method of second aspect, the determination unit is additionally operable to：According to service Level protocol SLA is that the management object in the large-scale cluster determines the multiple grade of service.

With reference to second aspect and its above-mentioned realization method, in second of realization method of second aspect, the determining list Member is additionally operable to：Determine the target capabilities of first service grade in the multiple grade of service；By the mesh of the first service grade Mark performance is determined as the target capabilities of at least one management object.

With reference to second aspect and its above-mentioned realization method, in the third realization method of second aspect, the determining list Member is specifically used for：The corresponding target capabilities of at least one management object are determined according to scheduled performance strategy；Or The target capabilities of at least one management object are manually set.

With reference to second aspect and its above-mentioned realization method, in the 4th kind of realization method of second aspect, the determining list The type of the determining target capabilities of member is included in response delay, read-write number IOPS per second, message transmission rate, CPU usage At least one.

It is described to obtain list in the 5th kind of realization method of second aspect with reference to second aspect and its above-mentioned realization method Member is specifically used for：Monitor periodically or routinely the actual performance of at least one management object.

With reference to second aspect and its above-mentioned realization method, in the 6th kind of realization method of second aspect, the performance pipe Reason unit is specifically used for：Whether the actual performance for determining to get by the determination unit meets the target capabilities； When the actual performance is unsatisfactory for the target capabilities, management object and/or described corresponding to the first service grade Except the corresponding management object of other grades of service of the first service grade carries out the performance management in multiple grades of service, So that the actual performance of the first service grade meets the target capabilities.

With reference to second aspect and its above-mentioned realization method, in the 7th kind of realization method of second aspect, the performance pipe Reason includes at least one of following：Business migration；Business limits；Flow control；Scheduling of resource；Send out alarm.

With reference to second aspect and its above-mentioned realization method, in the 8th kind of realization method of second aspect, in the reality When performance meets the target capabilities, the determination unit repeats the first service grade pair in multiple grades of service The step of at least one management object is determined in the management object answered or the acquiring unit are repeated described in the acquisition The step of actual performance of at least one management object.

With reference to second aspect and its above-mentioned realization method, in the 9th kind of realization method of second aspect, the determining list Member is specifically used for：

At least one management object for meeting predetermined condition is determined in the corresponding management object of the first service grade, Wherein described predetermined condition includes at least one of settling time, location information, loading condition and historical record；Or according to Pre-defined algorithm determines at least one management object in the corresponding management object of the first service grade, wherein the predetermined calculation Method is including randomly selecting, sequentially at least one of selection, time choice of dynamical.

With reference to second aspect and its above-mentioned realization method, in the 9th kind of realization method of second aspect, the management pair As including virtual machine VM, storage volume, virtual switch vSwitch, virtual local LAN vLAN, input and output I/O ports, handing over It changes planes, at least one of network bandwidth and server.

The embodiment of the present invention in the corresponding management object of the first service grade in large-scale cluster by determining at least one A management object, and it is corresponding to the first service grade according to the target capabilities and actual performance of at least one management object All management objects carry out performance management, and the performance so as to ensure most even whole users reaches Objective Can, improve or ensured user experience.

Description of the drawings

In order to illustrate the technical solution of the embodiments of the present invention more clearly, it will make below to required in the embodiment of the present invention Attached drawing is briefly described, it should be apparent that, drawings described below is only some embodiments of the present invention, for For those of ordinary skill in the art, without creative efforts, other are can also be obtained according to these attached drawings Attached drawing.

Fig. 1 is the system block diagram of the large-scale cluster management system of one embodiment of the invention；

Fig. 2 is the flow chart of the management method of one embodiment of the invention；

Fig. 3 is the flow chart of the management method of one embodiment of the invention；

Fig. 4 is the schematic block diagram of the managing device of one embodiment of the invention；

Fig. 5 is the schematic block diagram of the managing device of another embodiment of the present invention.

Specific embodiment

Below in conjunction with the attached drawing in the embodiment of the present invention, the technical solution in the embodiment of the present invention is carried out clear, complete Site preparation describes, it is clear that described embodiment is the part of the embodiment rather than whole embodiments of the present invention.Based on this hair Embodiment in bright, the every other reality that those of ordinary skill in the art are obtained under the premise of creative work is not made Example is applied, should all belong to the scope of protection of the invention.

Fig. 1 is the system block diagram of the management system of the large-scale cluster of one embodiment of the invention.The big rule shown in Fig. 1 The management system 100 of mould cluster includes：Object determining module 101, target capabilities determining module 102, actual performance is managed to obtain Module 103, performance management module 104 and large-scale cluster 105.Wherein management object determining module 101, actual performance obtain mould Block 103 and performance management module 104 are all connected with large-scale cluster 105, and management object determining module 101 is true with target capabilities Cover half block 102 is connected, target capabilities determining module 102 and actual performance acquisition module 103 all with 104 phase of performance management module Connection.

Object determining module 101 is managed to be used in the corresponding management object of first service grade of multiple grades of service really Fixed at least one management object, wherein management object is the resource unit in large-scale cluster 105.Resource unit can be divided into meter Calculate resource unit, storage resource cells, Internet resources unit, physical resource unit etc..More specifically, computing resource unit can Think virtual machine（Virtual Machine, VM）Deng storage resource cells can be storage volume and logical unit number（Logical Unit Number, LUN）Deng Internet resources unit can be input and output（Input/Output, I/O）Port, network bandwidth, Virtual switch（Virtual Switch,vSwitch）, virtual LAN (Virtual Local Area Network, VLAN), interchanger etc., physical resource unit can be server etc..

Target capabilities determining module 102 is used to determine the target capabilities of above-mentioned at least one management object, specifically, can be with The corresponding target capabilities of at least one management object are determined according to scheduled performance strategy；Or at least one management is manually set The target capabilities of object；Or above-mentioned at least one target capabilities for managing the corresponding first service grade of object are determined as this The target capabilities of at least one management object.

Actual performance acquisition module 103 is used to obtain the actual performance of above-mentioned at least one management object, specifically, can be with Monitor periodically or routinely and count the actual performance of at least one management object.

The target capabilities and actual performance that performance management module 104 is used to be determined according to target capabilities determining module 102 obtain The actual performance that modulus block 103 is got management object corresponding to first service grade carries out performance management.

Specifically, when actual performance is unsatisfactory for target capabilities, management object and/or more corresponding to first service grade Except the corresponding management object of other grades of service of first service grade carries out performance management in a grade of service, so that first The actual performance of the grade of service meets target capabilities, and the wherein method of performance management includes but not limited to following several：Business is moved It moves；Business limits；Flow control；Scheduling of resource；Send out alarm etc..

When actual performance meets target capabilities, can at least one pipe be redefined by target capabilities determining module 102 Reason object or the reality that can continue to monitor predetermined at least one management object by actual performance acquisition module 103 Performance.

The management system 100 of the large-scale cluster of the embodiment of the present invention passes through in the corresponding management object of first service grade In determine at least one management object, and according to this it is at least one management object target capabilities and actual performance to this first clothes The corresponding all management objects of grade of being engaged in carry out performance management, so as to ensure the performance of most even whole users Reach target capabilities, improve or ensured user experience.

Fig. 2 is the flow chart of the management method of one embodiment of the invention.

201, determine at least one management object in the corresponding management object of first service grade of multiple grades of service, Wherein management object is the resource unit in large-scale cluster.

202, determine the target capabilities of at least one management object.

203, obtain the actual performance of at least one management object.

204, performance management is carried out according to target capabilities and actual performance management object corresponding to first service grade.

The embodiment of the present invention in the corresponding management object of the first service grade in large-scale cluster by determining at least one A management object, and it is corresponding to the first service grade according to the target capabilities and actual performance of at least one management object All management objects carry out performance management, and the performance so as to ensure most even whole users reaches Objective Can, improve user experience.

It should be understood that the resource unit of large-scale cluster can be divided into computing resource unit, storage resource cells, Internet resources Unit, physical resource unit etc., for providing the services such as calculating, storage, transmission to the user.More specifically, computing resource list Member can be virtual machine VM etc., and storage resource cells can be storage volume and logical unit number LUN etc., and Internet resources unit can be with For input and output I/O ports, virtual switch vSwitch, virtual LAN vLAN, interchanger and network bandwidth etc., physics money Source unit can be server etc..

Optionally, it is true in the corresponding management object of first service grade of multiple grades of service as one embodiment Before fixed at least one management object, further include：According to service-level agreement（Service level Agreement, SLA）For Management object in large-scale cluster determines multiple grades of service.

It first, can be before management object be chosen first to the user in large-scale cluster as a preposition process Or management object carries out the division of the grade of service.Specifically can grade classification be carried out by SLA, can also be tieed up by network Shield personnel carry out grade classification according to certain attribute, such as location information, service type, service goal of management object etc.. When the object of grade classification is user, the object for being equal to grade classification is at least one resource list for the service that provides a user Member manages object.

In addition, the division of the grade of service can be simple grade classification, it can also be when carrying out grade of service division just Determine some/target capabilities of the multiple grades of service, here target capabilities can be understood as institute's service quality to be achieved （Quality of Service, QoS）.

Optionally, as one embodiment, multiple grades of service are determined for the management object in large-scale cluster according to SLA Later, it further includes：Determine the target capabilities of first service grade in multiple grades of service；Determine the mesh of at least one management object Performance is marked, including：The target capabilities of first service grade are determined as to the target capabilities of at least one management object.With reference to above-mentioned Embodiment, then can be by the grade of service in segmentation service grade if it have been determined that the target capabilities of the grade of service Target capabilities are determined as the target capabilities of at least one management object as sample chosen in the grade of service.

Optionally, as one embodiment, determine at least one management object target capabilities include it is following at least It is a kind of：The corresponding target capabilities of at least one management object are determined according to scheduled performance strategy；Or artificial setting at least one The target capabilities of a management object.

It, can also direct needle other than the above-mentioned service performance that the target capabilities of the grade of service are determined as to management object Its target capabilities is determined to determining at least one management object, can specifically be determined according to scheduled performance strategy, i.e., Performance strategy file can be preset in system, can determine by the certain attribute binding performance strategy files for managing object makes Object, which must be managed, can obtain the target capabilities of performance guarantee, and for example, strategy file can include the service of management object The correspondence of the information such as type, geographical location and target capabilities.Further, it is also possible to administration interface is passed through by network maintenance staff Manual setting manages the target capabilities of object.

Optionally, as one embodiment, the type of target capabilities can include but is not limited to response delay, read-write per second At least one of number IOPS, message transmission rate, CPU usage.It is readily appreciated that ground, target capabilities can be single ginseng The combination of number or many kinds of parameters, the present invention do not limit this.

Optionally, as one embodiment, the actual performance of at least one management object is obtained, including：Periodicity is held Monitor to continuous property the actual performance of at least one management object.It should be understood that actual performance can be identical with the type of target capabilities, It can also be different.

Optionally, as one embodiment, according to target capabilities and actual performance to the corresponding management of first service grade Object carries out performance management, including：Determine whether the actual performance got meets target capabilities；Mesh is unsatisfactory in actual performance Other of first service grade are removed when marking performance, in management object corresponding to first service grade and/or multiple grades of service The corresponding management object of the grade of service carries out performance management, so that the actual performance of first service grade meets target capabilities.

Optionally, performance management can include but is not limited at least one of following：Business migration；Business limits；Stream Amount control；Scheduling of resource；Send out alarm.

That is, if the actual performance detected is unsatisfactory for being expected（Target capabilities）, then can be to current detection First service grade or other grades of service carry out the operations such as business migration, business limitation, flow control, scheduling of resource and come So that the first service grade disclosure satisfy that target capabilities.For example, when at least one management pair selected in first service grade As the actual performance being monitored to is higher than 90% for CPU usage（Target capabilities are less than or equal to 90% for CPU usage）, then can be with Business migration is carried out to the management object of the first service grade, so that CPU usage is down to 90% or less, it should be appreciated that also The more moneys of management object distribution of target capabilities, for example, the first service grade can be reached using other regulation and control methods Source etc., the present invention do not limit this.

Furthermore, it is also possible to by the way that other grades of service are carried out with management and control or is dispatched to reach target come first service grade Performance, for example, when the actual performance I/O time delays of first service grade are unsatisfactory for target capabilities, it can be relatively low excellent by reducing The service traffics of the grade of service of first grade cause the first service grade to meet target capabilities.It is, of course, also possible to by right simultaneously First service grade and other grades of service carry out management and control or scheduling first service grade to be caused to reach target capabilities.In addition, It can be sent out alerting and management and control or scheduling wouldn't being carried out, wait for staff or the further instruction of other Network Management Equipments.No It loses in general manner, it can also be by carrying out performance management to first service grade, so that other grades of service reach expectation Energy.

Optionally, it when actual performance is unsatisfactory for target capabilities, can also repeat the first of multiple grades of service The step of determining at least one management object in the corresponding management object of the grade of service repeats at least one pipe of acquisition The step of managing the actual performance of object.It is detected or is continued for again to carry out that is, sampling can be re-started Monitoring.In this way, there can be higher essence by setting the threshold value of number of repetition come the sampling of performance management system and monitoring Degree, is more nearly practical situation.It is all unsatisfactory for for example, 2 actual performances monitored of repeated sampling can be preset Target capabilities, it is determined that carry out above-mentioned performance management.

Optionally, it as one embodiment, when actual performance meets target capabilities, repeats in multiple grades of service First service grade it is corresponding management object in determine it is at least one management object the step of or repeat acquisition at least One management object actual performance the step of.When performance satisfaction does not need to management and control or scheduling, resampling can be carried out, Select at least one management object again i.e. in first service grade.It can continue at least one pipe for prior sample Reason object is monitored, in order to carry out performance management when its performance is unsatisfactory for target capabilities.

Optionally, it is true in the corresponding management object of first service grade of multiple grades of service as one embodiment Fixed at least one management object, including：Determine to meet predetermined condition at least in the corresponding management object of first service grade One management object, wherein predetermined condition include at least one in settling time, location information, loading condition and historical record Kind；Or at least one management object is determined in the corresponding management object of first service grade according to pre-defined algorithm, wherein in advance Determine that algorithm includes randomly selecting, sequence is chosen, at least one of time choice of dynamical.

Optionally, as one embodiment, management object includes virtual machine VM, storage volume, input and output I/O ports, net At least one of network bandwidth and server.

Fig. 3 is the flow chart of the management method of one embodiment of the invention.

301, the grade of service divides

First, as an optional step, can choose manage object before to the user in large-scale cluster or Manage the division that object carries out the grade of service.Specifically can grade classification be carried out by SLA, it can also be by network operation people Member carries out grade classification according to certain attribute, such as location information, service type, service goal of management object etc..Wait When the object that grade divides is user, the object for being equal to grade classification is at least one resource unit for the service that provides a user, Manage object.

302, choose management object

A small amount of management object is chosen in large-scale cluster as management object, needs exist for ensureing in a grade of service At least one management object is chosen, wherein resource unit of the management object to provide service in large-scale cluster to the user.Specifically Ground, the resource unit of large-scale cluster can be divided into computing resource unit, storage resource cells, Internet resources unit, physics money Source unit etc., for providing the services such as calculating, storage, transmission to the user.More specifically, computing resource unit can be virtual Machine VM etc., storage resource cells can be storage volume and logical unit number LUN etc., and Internet resources unit can be input and output I/ O port and network bandwidth etc., physical resource unit can be server etc..

For first service grade, it can determine to meet predetermined item in the corresponding management object of first service grade At least one management object of part, wherein predetermined condition are included in settling time, location information, loading condition and historical record At least one, for example, predetermined condition, which reaches for loading condition in 90% or historical record of maximum load, n times failure occurred With first-class.It should be understood that at least one management object chosen, which can be same class, manages object, or inhomogeneous management Object for example, can all be VM, can also all be storage volume, can all be included, as long as they are to meet with VM, storage volume etc. Above-mentioned predetermined condition.Exist in addition, predetermined condition may be combining form, such as loading condition reaches maximum load 90% VM occurred more than n times failure server, etc. in historical record, and the present invention does not limit this.

Further, it is also possible at least one management is determined in the corresponding management object of first service grade according to pre-defined algorithm Object, wherein pre-defined algorithm include but not limited to randomly select, sequence selections, time choice of dynamical, intelligence selection etc..As one A example, if pre-defined algorithm to randomly select, when managing object select, is selected certain at random in first service grade The management object of quantity, quantity here equally can be preassigned in pre-defined algorithm, in another example, time choice of dynamical, Section or management object dynamically can be chosen with the variation of time in different times, can ensure sample in this way Activity.

Without loss of generality, the management object sampled can also be directly specified, such as can be existed by network maintenance staff In network topology interface one or more management objects, the sample as performance management are chosen for some grade of service.

It should be understood that since above-mentioned steps 301 are optional step, when step 301 performs, first in step 302 The grade of service is one in the multiple grades of service divided in above-mentioned steps 301, and herein, " first " grade of service is only used for It represents some grade of service, can be any one in above-mentioned multiple grades of service.When step 301 does not perform, on a large scale Still there may be the grade of service in cluster, which can be the grade of service that history determines or user signs The grade of service arranged when about networking, does not limit herein.The grade of service can be understood as will according to same or similar performance It asks, the management object grouping that performance indicator, type of service etc. are determining.

303, determine target capabilities

After determining as at least one management object of performance management sample, it may be determined that manage the Objective of object Energy.Specifically, the corresponding target capabilities of at least one management object can be determined according to scheduled performance strategy, it can also be artificial The target capabilities of at least one management object are set.That is, performance strategy file can be preset in system, pass through management Certain attribute binding performance strategy files of object, which can determine, enables management object to obtain the target capabilities of performance guarantee, For example, strategy file can include the information such as service type, the geographical location of management object pass corresponding with target capabilities System.Further, it is also possible to the target capabilities of object are managed by administration interface manual setting by network maintenance staff.For example, management Object is storage volume, has multiple grades of service, can be with for the storage volume that sample is chosen in one of grade of service Its target capabilities is set as time delay less than 3ms, which can be determined by manually setting or by strategy file 's.

In addition, it is also possible to which the grade of service has corresponded to target capabilities in advance（Service quality QoS）, for example, above-mentioned It, then can be by the grade of service if it have been determined that the target capabilities of the grade of service during segmentation service grade in step 301 Target capabilities are determined as the target capabilities of at least one management object as sample chosen in the grade of service.

The type of target capabilities has very much, can include but is not limited to response delay, read-write number IOPS per second, data and passes Defeated rate, CPU usage etc..It is readily appreciated that ground, target capabilities can be single parameter or the group of many kinds of parameters It closes, the present invention does not limit this.

304, monitor actual performance

The actual performance of at least one management object periodically or routinely determined in monitoring step 303.Detection The type of actual performance can be identical with target type, can also be different.Specifically, the Objective determined in above-mentioned steps 303 When can be less than 3ms for time delay, the type of the actual performance of detection may also be time delay, such as monitor the practical time delay of management object For 4ms.In addition, the actual performance situation different from target type of detection is also likely to be present, for example, target performance requirement is VM Creation time is less than 2min, and the actual performance index monitored is MBPS（Bandwidth）, then system think that 50MB/S is not achieved in MBPS, The goal nonreachable completed is created in VM2min into therefore carrying out performance strategy scheduling etc..

305, judge

System, can be to the data combining target performance of the actual performance detected after the actual performance of detection is received It is analyzed, that is, judges whether actual performance reaches target capabilities.That is, can by above-mentioned steps 302 determine The performance of the management object of sampling estimates decision entirely with the management object or cluster resource of the grade of service, in order to right The grade of service carries out total evaluation and management.

306, it is unsatisfactory for target capabilities

If by judging to determine that above-mentioned actual performance is unsatisfactory for target capabilities, it needs to be determined that carrying out the property of which kind of mode It can management.In general there are several performance management modes：Such as migration, limitation, scheduling, alarm etc..For example, target capabilities are set IO delays, IOPS and CPU usage are determined, the actual performance CPU usage that actual monitoring arrives is exceeded, then can specify migration plan Slightly, business migration is performed, reduces the business load of the management object of the grade of service, to meet user experience index request, together When can balance system-wide load；If actual performance IO time delays are exceeded, scheduling of resource can be carried out, increases this service etc. The resource proportioning of grade, such as CPU, caching, can also be met by the service traffics for the grade of service for limiting lower priority The demand of this grade of service.Furthermore it is also possible to send out alarm and management and control or scheduling wouldn't be carried out, staff or other nets are waited for The further instruction of tube apparatus.Further, it is also possible to by carrying out performance management to first service grade come other services etc. The demand of grade is met.

In addition, when actual performance is unsatisfactory for target capabilities, the first clothes in multiple grades of service can also be repeated The step of determining at least one management object in the corresponding management object of business grade repeats at least one management of acquisition The step of actual performance of object.It is detected or is continued for again to be supervised that is, sampling can be re-started It surveys.In this way, can have higher precision by setting the threshold value of number of repetition come the sampling of performance management system and monitoring, It is more nearly practical situation.For example, all discontented foot-eye of 2 actual performances monitored of repeated sampling can be preset Performance, it is determined that carry out above-mentioned performance management.

307, meet target capabilities

It, can be with return to step 302 or can be with return to step 304 when actual performance meets target capabilities.That is When performance meets without management and control or scheduling, resampling can be carried out, i.e., is selected again in first service grade At least one management object.It can continue to be monitored at least one management object of prior sample, in order at it Performance is unsatisfactory for carrying out performance management during target capabilities.

Fig. 4 is the schematic block diagram of the managing device of one embodiment of the invention.Managing device 400 in Fig. 4 includes determining Unit 401, acquiring unit 402 and capability management unit 403.

Determination unit 401 determines at least one pipe in the corresponding management object of first service grade of multiple grades of service Object is managed, wherein management object is the resource unit in large-scale cluster；Determination unit 401 determines at least one management object Target capabilities；Acquiring unit 402 obtains the actual performance of at least one management object.Capability management unit 403 is according to Objective It can management object progress performance management corresponding to first service grade with actual performance.

The managing device 400 of the embodiment of the present invention passes through the corresponding management object of first service grade in large-scale cluster In determine at least one management object, and according to this it is at least one management object target capabilities and actual performance to this first clothes The corresponding all management objects of grade of being engaged in carry out performance management, so as to ensure the performance of most even whole users Reach target capabilities, improve user experience.

It should be understood that the resource unit of large-scale cluster can be divided into computing resource unit, storage resource cells, Internet resources Unit, physical resource unit etc., for providing the services such as calculating, storage, transmission to the user.More specifically, computing resource list Member can be virtual machine VM etc., and storage resource cells can be storage volume and logical unit number LUN etc., and Internet resources unit can be with For input and output I/O ports and network bandwidth etc., physical resource unit can be server etc..

It should also be understood that the determination unit 401 in the embodiment of the present invention can correspond to above-mentioned large-scale cluster shown in FIG. 1 Management object determining module 101 and target capabilities determining module 102 in management system 100；Acquiring unit 402 can correspond to Actual performance acquisition module 103 in above-mentioned large-scale cluster management system 100 shown in FIG. 1；Capability management unit 403 can be with Corresponding to the performance management module 104 in above-mentioned large-scale cluster management system 100 shown in FIG. 1.

Optionally, as one embodiment, determination unit 401 is according to service-level agreement（Service level Agreement, SLA）Multiple grades of service are determined for the management object in large-scale cluster.

First, as a preposition process, can by determination unit 401 choose manage object before first to advising greatly User or management object in mould cluster carry out the division of the grade of service.Specifically can grade classification be carried out by SLA, It can also be by network maintenance staff according to certain attribute, such as location information, service type, service goal of management object etc. Carry out grade classification.When the object of grade classification is user, the object for being equal to grade classification is to provide a user service At least one resource unit, that is, manage object.

Optionally, as one embodiment, multiple grades of service are determined for the management object in large-scale cluster according to SLA Later, determination unit 401 can be also used for determining the target capabilities of first service grade in multiple grades of service；Determine at least one The target capabilities of a management object, including：The target capabilities of first service grade are determined as to the mesh of at least one management object Mark performance.It, then can be in segmentation service grade if it have been determined that the target capabilities of the grade of service with reference to above-described embodiment The target capabilities of the grade of service are determined as the mesh of at least one management object as sample chosen in the grade of service Mark performance.

Optionally, as one embodiment, determination unit 401 can be also used for being determined at least according to scheduled performance strategy The corresponding target capabilities of one management object；Or the target capabilities of at least one management object are manually set.

Other than the above-mentioned service performance that the target capabilities of the grade of service are determined as to management object, determination unit 401 Its target capabilities can also be determined directly against determining at least one management object, it specifically can be according to scheduled performance plan It slightly determines, i.e., performance strategy file can be preset in system, by the certain attribute binding performance strategies text for managing object Part, which can determine, enables management object to obtain the target capabilities of performance guarantee, and for example, strategy file can include pipe Manage the correspondence of the information such as service type, the geographical location of object and target capabilities.Further, it is also possible to by network maintenance staff The target capabilities of object are managed by administration interface manual setting.

Optionally, as one embodiment, acquiring unit 402 is specifically used for periodically or routinely monitors at least one Manage the actual performance of object.It should be understood that actual performance can be identical with the type of target capabilities, can also be different.

Optionally, as one embodiment, capability management unit 403 is specifically used for whether determining the actual performance got Meet target capabilities；When actual performance is unsatisfactory for target capabilities, management object and/or multiple corresponding to first service grade Except the corresponding management object of other grades of service of first service grade carries out performance management in the grade of service, so that the first clothes The actual performance of business grade meets target capabilities.

Furthermore, it is also possible to by the way that other grades of service are carried out with management and control or is dispatched to reach target come first service grade Performance, for example, when the actual performance I/O time delays of first service grade are unsatisfactory for target capabilities, it can be relatively low excellent by reducing The service traffics of the grade of service of first grade cause the first service grade to meet target capabilities.It is, of course, also possible to by right simultaneously First service grade and other grades of service carry out management and control or scheduling first service grade to be caused to reach target capabilities.In addition, It can be sent out alerting and management and control or scheduling wouldn't being carried out, wait for staff or the further instruction of other Network Management Equipments.

Optionally, as one embodiment, when actual performance meets target capabilities, determination unit 401 repeats The step of determining at least one management object in the corresponding management object of first service grade of multiple grades of service obtains Unit 402 repeats the step of actual performance for obtaining at least one management object.When performance satisfaction do not need to management and control or During scheduling, resampling can be carried out, i.e., selectes at least one management object again in first service grade.It can continue to It is monitored at least one management object of prior sample, in order to carry out performance when its performance is unsatisfactory for target capabilities Management.

Optionally, as one embodiment, determination unit 401 is additionally operable in the corresponding management object of first service grade Determine at least one management object for meeting predetermined condition, wherein predetermined condition includes settling time, location information, loading condition At least one of with historical record；Or it is determined at least in the corresponding management object of first service grade according to pre-defined algorithm One management object, wherein pre-defined algorithm are including randomly selecting, sequentially at least one of selection, time choice of dynamical.

Optionally, as one embodiment, management object includes virtual machine VM, storage volume, input and output I/O ports, void Intend at least one of interchanger vSwitch, virtual LAN vLAN, interchanger, network bandwidth and server.

The managing device 400 of the embodiment of the present invention passes through the corresponding management object of first service grade in large-scale cluster In determine at least one management object, and according to this it is at least one management object target capabilities and actual performance to this first clothes The corresponding all management objects of grade of being engaged in carry out performance management, so as to ensure the performance of most even whole users Reach target capabilities, improve or ensured user experience.

Fig. 5 is the schematic block diagram of the managing device of another embodiment of the present invention.The managing device 500 of Fig. 5 includes processor 51 and memory 52, processor 51 be connected with memory 52 by bus system 53.

Memory 52 causes processor 51 to perform the instruction operated below for storing：In the first clothes of multiple grades of service At least one management object is determined in the corresponding management object of grade of being engaged in, wherein management object is the resource list in large-scale cluster Member；Determine the target capabilities of at least one management object；Obtain the actual performance of at least one management object；According to target capabilities Management object corresponding to first service grade with actual performance carries out performance management.

The managing device 500 of the embodiment of the present invention passes through the corresponding management object of first service grade in large-scale cluster In determine at least one management object, and according to this it is at least one management object target capabilities and actual performance to this first clothes The corresponding all management objects of grade of being engaged in carry out performance management, so as to ensure the performance of most even whole users Reach target capabilities, improve user experience.

In addition, managing device 50 can also include radiating circuit 54, receiving circuit 55 etc..Processor 51 controls managing device 50 operation, processor 51 can also be known as CPU（Central Processing Unit, central processing unit）.Memory 52 It can include read-only memory and random access memory, and instruction and data is provided to processor 51.One of memory 52 Nonvolatile RAM can also be included by dividing（NVRAM）.The various components of managing device 50 pass through bus system 53 It is coupled, wherein bus system 53 can also include power bus, controlling bus and state in addition to including data/address bus Signal bus etc..But for the sake of clear explanation, various buses are all designated as bus system 53 in figure.

The method that the embodiments of the present invention disclose can be applied to realize in processor 51 or by processor 51.Place It may be a kind of IC chip to manage device 51, has the processing capacity of signal.During realization, each step of the above method It can be completed by the integrated logic circuit of the hardware in processor 51 or the instruction of software form.Above-mentioned processor 51 can To be general processor, digital signal processor（DSP）, application-specific integrated circuit（ASIC）, ready-made programmable gate array（FPGA） Either other programmable logic device, discrete gate or transistor logic, discrete hardware components.It can realize or perform Disclosed each method, step and logic diagram in the embodiment of the present invention.General processor can be microprocessor or this at It can also be any conventional processor etc. to manage device.The step of method with reference to disclosed in the embodiment of the present invention, can directly embody Completion is performed for hardware decoding processor or performs completion with the hardware in decoding processor and software module combination.Software Module can be located at random access memory, flash memory, read-only memory, programmable read only memory or electrically erasable programmable storage In the storage medium of this fields such as device, register maturation.The storage medium is located at memory 52, and processor 51 reads memory 52 In information, with reference to its hardware complete the above method the step of.

Optionally, it is true in the corresponding management object of first service grade of multiple grades of service as one embodiment Before fixed at least one management object, further include：It is determined according to service-level agreement SLA for the management object in large-scale cluster Multiple grades of service.

Optionally, as one embodiment, multiple grades of service are determined for the management object in large-scale cluster according to SLA Later, it further includes：Determine the target capabilities of first service grade in multiple grades of service；Determine the mesh of at least one management object Performance is marked, including：The target capabilities of first service grade are determined as to the target capabilities of at least one management object.

Optionally, as one embodiment, the type of target capabilities includes response delay, read-write number IOPS per second, number According at least one of transmission rate, CPU usage.

Optionally, as one embodiment, the actual performance of at least one management object is obtained, including：Periodicity is held Monitor to continuous property the actual performance of at least one management object.

Optionally, as one embodiment, performance management includes at least one of following：Business migration；Business limits； Flow control；Scheduling of resource；Send out alarm.

Optionally, it as one embodiment, when actual performance meets target capabilities, repeats in multiple grades of service First service grade it is corresponding management object in determine it is at least one management object the step of or repeat acquisition at least One management object actual performance the step of.

Optionally, as one embodiment, management object includes virtual machine VM, storage volume, input and output I/O ports, net At least one of network bandwidth, virtual switch vSwitch, virtual LAN vLAN, interchanger and server.

The managing device 500 of the embodiment of the present invention passes through the corresponding management object of first service grade in large-scale cluster In determine at least one management object, and according to this it is at least one management object target capabilities and actual performance to this first clothes The corresponding all management objects of grade of being engaged in carry out performance management, so as to ensure the performance of most even whole users Reach target capabilities, improve or ensured user experience.

It should be understood that the terms "and/or", only a kind of incidence relation for describing affiliated partner, expression can deposit In three kinds of relationships, for example, A and/or B, can represent：Individualism A exists simultaneously A and B, these three situations of individualism B. In addition, character "/" herein, it is a kind of relationship of "or" to typically represent forward-backward correlation object.

It should be understood that in various embodiments of the present invention, the size of the serial number of above-mentioned each process is not meant to perform suitable The priority of sequence, the execution sequence of each process should be determined with its function and internal logic, without the implementation of the reply embodiment of the present invention Process forms any restriction.

Those of ordinary skill in the art may realize that each exemplary lists described with reference to the embodiments described herein Member and algorithm steps can be realized with the combination of electronic hardware or computer software and electronic hardware.These functions are actually It is performed with hardware or software mode, specific application and design constraint depending on technical solution.Professional technician Described function can be realized using distinct methods to each specific application, but this realization is it is not considered that exceed The scope of the present invention.

It is apparent to those skilled in the art that for convenience and simplicity of description, the system of foregoing description, The specific work process of device and unit can refer to the corresponding process in preceding method embodiment, and details are not described herein.

In several embodiments provided herein, it should be understood that disclosed systems, devices and methods, it can be with It realizes by another way.For example, the apparatus embodiments described above are merely exemplary, for example, the unit It divides, only a kind of division of logic function can have other dividing mode, such as multiple units or component in actual implementation It may be combined or can be integrated into another system or some features can be ignored or does not perform.Another point, it is shown or The mutual coupling, direct-coupling or communication connection discussed can be the indirect coupling by some interfaces, device or unit It closes or communicates to connect, can be electrical, machinery or other forms.

The unit illustrated as separating component may or may not be physically separate, be shown as unit The component shown may or may not be physical unit, you can be located at a place or can also be distributed to multiple In network element.Some or all of unit therein can be selected according to the actual needs to realize the mesh of this embodiment scheme 's.

In addition, each functional unit in each embodiment of the present invention can be integrated in a processing unit, it can also That each unit is individually physically present, can also two or more units integrate in a unit.

If the function is realized in the form of SFU software functional unit and is independent product sale or in use, can be with It is stored in a computer read/write memory medium.Based on such understanding, technical scheme of the present invention is substantially in other words The part contribute to the prior art or the part of the technical solution can be embodied in the form of software product, the meter Calculation machine software product is stored in a storage medium, is used including some instructions so that a computer equipment（Can be People's computer, server or network equipment etc.）Perform all or part of the steps of the method according to each embodiment of the present invention. And aforementioned storage medium includes：USB flash disk, mobile hard disk, read-only memory（ROM, Read-Only Memory）, arbitrary access deposits Reservoir（RAM, Random Access Memory）, the various media that can store program code such as magnetic disc or CD.

The above description is merely a specific embodiment, but protection scope of the present invention is not limited thereto, any Those familiar with the art in the technical scope disclosed by the present invention, can readily occur in change or replacement, should all contain Lid is within protection scope of the present invention.Therefore, protection scope of the present invention should be based on the protection scope of the described claims.

Claims

1. a kind of management method of large-scale cluster, which is characterized in that including：

At least one management object is determined in the corresponding management object of first service grade of multiple grades of service, wherein described Object is managed as the resource unit in the large-scale cluster；

Determine the target capabilities of at least one management object；

Obtain the actual performance of at least one management object；

Performance pipe is carried out according to the target capabilities and the actual performance management object corresponding to the first service grade Reason,

Wherein, at least one management pair is determined in the corresponding management object of the first service grade in multiple grades of service As, including：

At least one management object for meeting predetermined condition is determined in the corresponding management object of the first service grade, wherein The predetermined condition includes at least one of settling time, location information, loading condition and historical record；Or

At least one management object, wherein institute are determined in the corresponding management object of the first service grade according to pre-defined algorithm State that pre-defined algorithm includes randomly selecting, sequence is chosen, at least one of time choice of dynamical.

2. according to the method described in claim 1, it is characterized in that, the first service grade in multiple grades of service corresponds to Management object in determine it is at least one management object before, further include：It is the extensive collection according to service-level agreement SLA Management object in group determines the multiple grade of service.

3. according to the method described in claim 2, it is characterized in that, it is described according to SLA be the large-scale cluster in management After object determines multiple grades of service, further include：Determine the target capabilities of first service grade in the multiple grade of service；

The target capabilities for determining at least one management object, including：By the target capabilities of the first service grade It is determined as the target capabilities of at least one management object.

4. according to the method in claim 2 or 3, which is characterized in that the mesh for determining at least one management object Performance is marked including at least one of following：The corresponding institute of at least one management object is determined according to scheduled performance strategy State target capabilities；Or the target capabilities of at least one management object are manually set.

5. according to the method described in any one in claim 1-3, which is characterized in that the type of the target capabilities includes ringing Answer at least one of time delay, read-write number IOPS per second, message transmission rate, CPU usage.

6. the according to the method described in claim 5, it is characterized in that, practical property for obtaining at least one management object Can, including：Monitor periodically or routinely the actual performance of at least one management object.

It is 7. according to the method described in claim 1, it is characterized in that, described according to the target capabilities and the actual performance pair The corresponding object that manages of the first service grade carries out performance management, including：

Whether the actual performance for determining to get meets the target capabilities；

When the actual performance is unsatisfactory for the target capabilities, it is corresponding to the first service grade management object and/or Except the corresponding management object of other grades of service of the first service grade carries out the performance in the multiple grade of service Management, so that the actual performance of the first service grade meets the target capabilities.

8. the method according to the description of claim 7 is characterized in that the performance management is including at least one of following：Industry Business migration；Business limits；Flow control；Scheduling of resource；Send out alarm.

9. the method according to the description of claim 7 is characterized in that when the actual performance meets the target capabilities, weigh Multiple perform determines at least one management object in the corresponding management object of the first service grade in multiple grades of service Step or repeat it is described obtain it is described it is at least one management object actual performance the step of.

10. according to the method described in any one in claim 1-3, which is characterized in that the management object includes virtual machine VM, storage volume, virtual switch vSwitch, virtual local LAN vLAN, input and output I/O ports, network bandwidth, exchange At least one of machine and server.

11. a kind of managing device of large-scale cluster, which is characterized in that including：

Determination unit, for determining at least one management in the corresponding management object of the first service grade of multiple grades of service Object, wherein the management object is the resource unit in the large-scale cluster；

The determination unit is additionally operable to determine the target capabilities of at least one management object；

Acquiring unit, for obtaining the actual performance of at least one management object；

Capability management unit, for according to the target capabilities and the actual performance to the corresponding pipe of the first service grade It manages object and carries out performance management,

Wherein, the determination unit is specifically used for：

12. according to the devices described in claim 11, which is characterized in that the determination unit is additionally operable to：It is assisted according to the grade of service View SLA is that the management object in the large-scale cluster determines the multiple grade of service.

13. device according to claim 12, which is characterized in that the determination unit is additionally operable to：

Determine the target capabilities of first service grade in the multiple grade of service；

The target capabilities of the first service grade are determined as to the target capabilities of at least one management object.

14. device according to claim 12 or 13, which is characterized in that the determination unit is specifically used for：According to predetermined Performance strategy determine the corresponding target capabilities of at least one management object；Or it manually sets described at least one Manage the target capabilities of object.

15. according to the device described in any one in claim 11-13, which is characterized in that the mesh that the determination unit determines The type for marking performance includes at least one of response delay, read-write number IOPS per second, message transmission rate, CPU usage.

16. device according to claim 15, which is characterized in that the acquiring unit is specifically used for：Periodically or continue Property monitor it is described it is at least one management object actual performance.

17. according to the devices described in claim 11, which is characterized in that the capability management unit is specifically used for：

Whether the actual performance for determining to get by the determination unit meets the target capabilities；

18. device according to claim 17, which is characterized in that the performance management includes at least one of following： Business migration；Business limits；Flow control；Scheduling of resource；Send out alarm.

19. device according to claim 17, which is characterized in that when the actual performance meets the target capabilities, The determination unit is repeated to be determined at least in the corresponding management object of the first service grade in multiple grades of service The step of one management object or the acquiring unit repeat the reality for obtaining at least one management object The step of performance.

20. according to the device described in any one in claim 11-13, which is characterized in that the management object includes virtual Machine VM, storage volume, virtual switch vSwitch, virtual local LAN vLAN, input and output I/O ports, interchanger, network At least one of bandwidth and server.