CN103778474A - Resource load capacity prediction method, analysis prediction system and service operation monitoring system - Google Patents

Abstract

The invention provides a resource load capacity prediction method, an analysis prediction system and a service operation monitoring system. The method comprises the following steps: respectively obtaining a service element historical data sequence and a resource load capacity historical data sequence; according to the service element historical data sequence, generating an expectation service element; according to the service element historical data sequence and the resource load capacity historical data sequence, establishing a correlation function between a service element and a resource load capacity; and according to the correlation function, generating an expectation resource load capacity corresponding to the expectation service element. In the embodiments of the invention, the application disposition demands of different users and the load balancing characteristics of various applications are fully considered, corresponding resources are disposed for the different users, the resource load capacities needed by the different applications in the future are accurately predicted according to the load balancing characteristics of the different applications, the resource disposition and maintenance mechanism of a cloud computing platform is improved, and the utilization rate of infrastructure resources of a cloud computing system is improved.

Description

Resource load amount Forecasting Methodology, analysing and predicting system and service operation supervisory system

Technical field

The present invention relates to the communication technology and network technology, relate in particular to a kind of resource load amount Forecasting Methodology, analysing and predicting system and service operation supervisory system.

Background technology

Along with the maturation of cloud, the charge capacity (as the cpu busy percentage of virtual machine and memory source amount) that cloud computing industry trends towards for basic resource is analyzed and predicts, has occurred cloud computing system framework as shown in Figure 1.As shown in Figure 1, this cloud computing system framework comprises: service management system, basic resource management system and analysing and predicting system.Wherein, service management system is mainly realized SLA(Service-Level Agreement, service-level agreement) function such as management, service catalogue management, accounting management, customer account management.Basic resource management system mainly realizes the functions such as monitoring resource, resource control, dynamic resource allocation, resource dynamic migration.Analysing and predicting system is mainly realized the functions such as data analysis, trend analysis, prediction alarm.Data analysis function is mainly that the basic resource charge capacity to gathering is carried out pre-service; Trend analysis function is mainly for the basic resource charge capacity after analyzing, according to specific forecast model and prediction algorithm, the variation tendency of certain time basis resource load amount of predict future; Prediction alarm function, mainly for the abnormality resource data of trend prediction, sends to service management system in the mode of network service, informs each user by service management system with the form of note or mail.

In prior art, the single resource load amount monitoring according to described basic resource management system of described analysing and predicting system is as historical data, to the prediction of Future charge capacity, do not consider the impact of other extraneous factors such as load characteristic on resource load amount of each application in cloud computing platform, prediction accuracy is low.

Summary of the invention

Many aspects of the present invention provide a kind of resource load amount Forecasting Methodology, analysing and predicting system and service operation supervisory system, in order to improve the accuracy of prediction.

First aspect of the present invention, provides a kind of resource load amount Forecasting Methodology, comprising:

Obtain respectively business factor historical data sequence and resource load amount historical data sequence, wherein, the monitoring time that in described business factor historical data sequence, each business factor information is carried is identical with the monitoring time that each resource load amount information in described resource load amount historical data sequence is carried respectively, described business factor information is that the business factor monitoring and monitoring time are together stored to the information in database by service operation supervisory system, described resource load amount information is that the resource load amount monitoring and monitoring time are together stored to the information in described database by resource management system,

According to described business factor historical data sequence, generate expection business factor;

According to described business factor historical data sequence and resource load amount historical data sequence, set up the correlation function of business factor and resource load amount;

According to described correlation function, generate the expection resource load amount that described expection business factor is corresponding.

Resource load amount Forecasting Methodology as above, wherein, described obtain respectively business factor historical data sequence and resource load amount historical data sequence before, also comprise:

Obtain business factor information;

Judge that monitoring time that described business factor information carries is whether within the default unusual day period, if, described business factor information is deposited in to unusual day historical data sequence of business factor, otherwise, deposit described business factor information in business factor normal day historical data sequence;

Gains resources charge capacity information;

Judge that monitoring time that described resource load amount information carries is whether within the default unusual day period, if, described resource load amount information is deposited in to unusual day historical data sequence of resource load amount, otherwise, deposit described resource load amount information in resource load amount normal day historical data sequence;

Correspondingly, described obtaining respectively in business factor historical data sequence and resource load amount historical data sequence, described business factor historical data sequence is unusual day historical data sequence of business factor, described resource load amount historical data sequence is unusual day historical data sequence of resource load amount, or described business factor historical data sequence is the normal day historical data sequence of business factor, described resource load amount historical data sequence is the normal day historical data sequence of resource load amount.

Resource load amount Forecasting Methodology as above, wherein, described obtain respectively business factor historical data sequence and resource load amount historical data sequence after, also comprise:

Respectively described business factor historical data sequence and resource load amount historical data sequence are carried out to data pre-service.

Resource load amount Forecasting Methodology as above, wherein, described according to described correlation function, after calculating the described expection resource load amount that described expection business factor is corresponding, also comprise:

Described expection resource load amount is stored in database, so that described service management system reads described expection resource load amount from described database, and generating resource deployment instruction according to described expection resource load amount, described resource management system is carried out resource deployment according to described resource deployment instruction.

Resource load amount Forecasting Methodology as above, wherein, described according to described business factor historical data sequence, generate expection business factor, comprising:

According to described business factor historical data sequence, draw the predicated error of at least two kinds of forecast models;

According to described business factor historical data sequence, adopt the forecast model of described predicated error minimum to calculate expection business factor.

Resource load amount Forecasting Methodology as above, wherein, described according to described business factor historical data sequence and resource load amount historical data sequence, set up the correlation function of business factor and resource load amount, comprising:

According to described business factor historical data sequence and resource load amount historical data sequence, adopt at least two function models to carry out the Function Fitting processing of described business factor and resource load amount, generate at least two candidate association functions;

According to described business factor historical data sequence and resource load amount historical data sequence, calculate the matching residual error that adopts each candidate association function according to default least-squares calculation algorithm;

The matching residual error of more each candidate association function, using the candidate association function of matching residual error minimum as described business factor and the correlation function of resource load amount.

Second aspect of the present invention, provides a kind of resource load amount Forecasting Methodology, comprising:

Monitoring business factor;

The described business factor monitoring and monitoring time are stored in database, so that analysing and predicting system obtains business factor historical data sequence from described database, according to described business factor historical data sequence, generate expection business factor, according to described business factor historical data sequence and resource load amount historical data sequence, set up the correlation function of business factor and resource load amount, according to described correlation function, generate the expection resource load amount that described expection business factor is corresponding;

Wherein, the monitoring time that in described business factor historical data sequence, each business factor information is carried is identical with the monitoring time that each resource load amount information in described resource load amount historical data sequence is carried respectively, described resource load amount historical data sequence is obtained from described database by described analysing and predicting system, and described resource load amount information is that the resource load amount monitoring and monitoring time are together stored to the information in described database by resource management system.

Resource load amount Forecasting Methodology as above, before described monitoring business factor, also comprises:

Receive the Monitoring instruction information that service management system sends, described Monitoring instruction information carries pre-monitoring business factor; Correspondingly

Described monitoring business factor, is specially:

According to described Monitoring instruction information, monitoring business factor.

The 3rd aspect of the present invention, provides a kind of analysing and predicting system, comprising:

The first acquisition module, for obtaining respectively business factor historical data sequence and resource load amount historical data sequence, wherein, the monitoring time that in described business factor historical data sequence, each business factor information is carried is identical with the monitoring time that each resource load amount information in described resource load amount historical data sequence is carried respectively, described business factor information is that the business factor monitoring and monitoring time are together stored to the information in database by service operation supervisory system, described resource load amount information is that the resource load amount monitoring and monitoring time are together stored to the information in described database by resource management system,

The first processing module, for according to described business factor historical data sequence, generates expection business factor;

Set up module, for according to described business factor historical data sequence and resource load amount historical data sequence, set up the correlation function of business factor and resource load amount;

The second processing module, for according to described correlation function, generates the expection resource load amount that described expection business factor is corresponding.

Analysing and predicting system as above, also comprises:

The second acquisition module, for obtaining business factor information;

Judge module, for judging that monitoring time that described business factor information carries whether within the default unusual day period, if so, generates the first judgement information, otherwise, generate the second judgement information;

Execution module, for according to described the first judgement information, deposits described business factor information in business factor unusual day historical data sequence; According to described the second judgement information, deposit described business factor information in business factor normal day historical data sequence;

Described the second acquisition module, also for Gains resources charge capacity information;

Described judge module, also for judging that monitoring time that described resource load amount information carries whether within the default unusual day period, if so, generates the 3rd judgement information, otherwise, generate the 4th judgement information;

Described execution module, also, for according to described the 3rd judgement information, deposits described resource load amount information in resource load amount unusual day historical data sequence; According to described the second judgement information, deposit described resource load amount information in resource load amount normal day historical data sequence;

Correspondingly, the described business factor historical data sequence that described the first acquiring unit obtains is unusual day historical data sequence of business factor, described resource load amount historical data sequence is unusual day historical data sequence of resource load amount, or, the described business factor historical data sequence that described the first acquiring unit obtains is the normal day historical data sequence of business factor, and described resource load amount historical data sequence is the normal day historical data sequence of resource load amount.

Analysing and predicting system as above, also comprises:

Data preprocessing module, carries out data pre-service for the historical data of the historical data to described business factor and described resource load amount respectively.

Analysing and predicting system as above, also comprises:

Memory module, for described expection resource load amount is stored to database, so that described service management system reads described expection resource load amount from described database, and generating resource deployment instruction according to described expection resource load amount, described resource management system is carried out resource deployment according to described resource deployment instruction.

Analysing and predicting system as above, described the first processing module, comprising:

The first processing unit, for according to described business factor historical data sequence, draws the predicated error of at least two kinds of forecast models;

The second processing unit, for according to described business factor historical data sequence, adopts the forecast model of described predicated error minimum to calculate expection business factor.

Analysing and predicting system as above, the described module of setting up comprises:

Matching unit, for according to described business factor historical data sequence and resource load amount historical data sequence, adopts at least two function models to carry out the Function Fitting processing of described business factor and resource load amount, generates at least two candidate association functions;

The 3rd processing unit, for according to described business factor historical data sequence and resource load amount historical data sequence, calculates the matching residual error that adopts each candidate association function according to default least-squares calculation algorithm;

Comparing unit, for the matching residual error of more each candidate association function, using the candidate association function of matching residual error minimum as described business factor and the correlation function of resource load amount.

The 4th aspect of the present invention, provides a kind of service operation supervisory system, comprising:

Monitoring modular, for monitoring business factor;

Memory module, for the described business factor and the monitoring time that monitor are stored to database, so that analysing and predicting system obtains business factor historical data sequence from described database, according to described business factor historical data sequence, generate expection business factor, according to described business factor historical data sequence and resource load amount historical data sequence, set up the correlation function of business factor and resource load amount, according to described correlation function, generate the expection resource load amount that described expection business factor is corresponding;

Wherein, the monitoring time that in described business factor historical data sequence, each business factor information is carried is identical with the monitoring time that each resource load amount information in described resource load amount historical data sequence is carried respectively, described resource load amount historical data sequence is obtained from described database by described analysing and predicting system, and described resource load amount information is that the resource load amount monitoring and monitoring time are together stored to the information in described database by resource management system.

Service operation supervisory system as above, also comprises:

Receiver module, the Monitoring instruction information sending for receiving service management system, described Monitoring instruction information carries pre-monitoring business factor, correspondingly

Described monitoring modular specifically for, according to described Monitoring instruction information, monitoring business factor.

The 5th aspect of the present invention, provides a kind of cloud computing system framework, comprising: service management system, database, resource management system, above-mentioned analysing and predicting system, and above-mentioned service operation supervisory system.

As shown from the above technical solution, the embodiment of the present invention, by introducing business factor as influence factor in Forecasting Methodology, by the correlation function between matching business factor and resource load amount, has realized the Accurate Prediction of the resource load amount to-be to this application.The embodiment of the present invention has taken into full account the application deployment demand of different user and the load balancing characteristic of each application, for different user is disposed corresponding resource, dope accurately the following required resource load amount of different application according to the load balancing characteristic of different application, improve resource deployment and the maintenance mechanism of cloud computing platform, contributed to the utilization factor of the infrastructure resources that improves cloud computing system.

Accompanying drawing explanation

In order to be illustrated more clearly in the embodiment of the present invention or technical scheme of the prior art, to the accompanying drawing of required use in embodiment or description of the Prior Art be briefly described below, apparently, accompanying drawing in the following describes is some embodiments of the present invention, for those of ordinary skills, do not paying under the prerequisite of creative work, can also obtain according to these accompanying drawings other accompanying drawing.

Fig. 1 is the structural representation of prior art cloud computing system framework;

Fig. 2 is the process flow diagram of charge capacity Forecasting Methodology embodiment mono-provided by the invention;

Fig. 3 is the process flow diagram of charge capacity Forecasting Methodology embodiment bis-provided by the invention;

Fig. 4 is the process flow diagram of charge capacity Forecasting Methodology embodiment tri-provided by the invention;

Fig. 5 is the curve synoptic diagram of an instantiation of the business factor historical data sequence of GPS application;

Fig. 6 is the curve synoptic diagram of an instantiation of the resource load amount historical data sequence of GPS application;

The chart schematic diagram of the expection business factor that Fig. 7 is the GPS application on July 1,1 day～2012 January in 2012 that the business factor historical data sequence prediction of the charge capacity Forecasting Methodology that provides of the embodiment of the present invention based on shown in Fig. 5 is provided;

Fig. 8 is the principle schematic of digital simulation residual error;

The chart schematic diagram of the expection resource load amount of the GPS application on July 1,1 day～2012 January in 2012 that the expection business factor that Fig. 9 applies based on the GPS shown in Fig. 7 for the charge capacity Forecasting Methodology that the employing embodiment of the present invention provides draws according to the correlation function calculating;

Figure 10 is the structural representation of analysing and predicting system embodiment mono-provided by the invention;

Figure 11 is the structural representation of analysing and predicting system embodiment bis-provided by the invention;

Figure 12 is the structural representation of an instantiation of the first processing module in analysing and predicting system embodiment provided by the invention;

Figure 13 is the structural representation of setting up an instantiation of module in analysing and predicting system embodiment provided by the invention;

Figure 14 is the structural representation of service operation supervisory system embodiment mono-provided by the invention;

Figure 15 is the structural representation of cloud computing system framework embodiment mono-provided by the invention.

Embodiment

For making object, technical scheme and the advantage of the embodiment of the present invention clearer, below in conjunction with the accompanying drawing in the embodiment of the present invention, technical scheme in the embodiment of the present invention is clearly and completely described, obviously, described embodiment is the present invention's part embodiment, rather than whole embodiment.Based on the embodiment in the present invention, those of ordinary skills, not making the every other embodiment obtaining under creative work prerequisite, belong to the scope of protection of the invention.

Fig. 2 is the process flow diagram of resource load amount Forecasting Methodology embodiment mono-provided by the invention, and as shown in Figure 2, the method comprises:

Step 101, obtain respectively business factor historical data sequence and resource load amount historical data sequence, wherein, the monitoring time that in described business factor historical data sequence, each business factor information is carried is identical with the monitoring time that each resource load amount information in described resource load amount historical data sequence is carried respectively, described business factor information is that the business factor monitoring and monitoring time are together stored to the information in database by service operation supervisory system, described resource load amount information is that the resource load amount monitoring and monitoring time are together stored to the information in described database by resource management system.

Particularly, analysing and predicting system obtains respectively business factor historical data sequence and resource load amount historical data sequence.Wherein, the each business factor information in described business factor historical data sequence should be corresponding with each resource load amount information in resource load amount historical data sequence respectively.Be service operation supervisory system while monitoring business factor, described resource management system should be monitored the resource load amount that now business factor is corresponding simultaneously.Therefore, described service operation supervisory system, in the time monitoring described business factor, needs to record monitoring time simultaneously, and described business factor is stored in data together with monitoring time.Described resource management system, in the time monitoring described resource load amount, similarly needs to record monitoring time, and described resource load amount is stored in data together with monitoring time.Particularly, in order to guarantee business factor and the correspondence of resource load amount on monitoring time, described service operation supervisory system is monitored business factor according to Preset Time or predetermined period, and the business factor monitoring and monitoring time are together stored in database.Similarly, described resource management system is monitored resource load amount according to the Preset Time the same with described service operation supervisory system or predetermined period, and the resource load amount monitoring and monitoring time are together stored in database.Wherein, described business factor can be specifically: registered user's number, while online user number, simultaneously online number of request etc.For example, GPS service application can be divided into three types, is respectively category-A GPS service application, category-B GPS service application and C class GPS service application.Wherein, category-A GPS service application, according to user scope, as number segment, is carried out relative static load balancing in advance, increases the granularity of division of number segment in the time of overload, and therefore category-A GPS service application Main Basis registered user number carries out load balancing.Category-B GPS service application is asked relative dynamic Resources allocation according to user, but owing to having incidence relation between such applied business request, the request of unique user can be bundled on specific application example within a certain period of time, therefore category-B GPS service application Main Basis simultaneously online user number carry out load balancing and resource deployment.C class GPS service application is distributed according to user's request dynamic, owing to there is no relevance between such application request, can will ask to be arbitrarily routed to resource node arbitrarily, therefore the simultaneously online number of request of C class GPS service application Main Basis is carried out the dynamic application of resource.Described resource load amount can comprise following one or more: CPU usage, internal memory use amount, hard disk use amount and network bandwidth occupancy.

Step 102, according to described business factor historical data sequence, generate expection business factor.

Particularly, analysing and predicting system, first, according to described business factor historical data sequence, draws the predicated error of at least two kinds of default forecast models; Then,, according to described business factor historical data sequence, adopt the forecast model of described predicated error minimum to calculate expection business factor.Wherein, according to described business factor historical data sequence, draw the predicated error of at least two kinds of default forecast models, specifically can adopt with the following method and realize.The method comprises: the monitoring information that the business factor information in described business factor historical data sequence is carried according to business factor is divided, and for example, business factor historical data sequence includes the business factor information of August 1 to September 5.The temporal information of carrying according to business factor information, divides the business factor information in sequence, is divided into two groups, and if August 1 was the first array to the business factor on August 31, September 1 to the business factor on September 5 is the second array.Then, using the first array as historical data, the second array is as predicted data, adopts forecast model, calculates respectively and the predicted data of the corresponding monitoring time of each business factor in the second array according to the first array.Finally, respectively the business factor of comparison prediction data and each corresponding monitoring time in the second array, draws difference, calculates the predicated error of this forecast model according to each difference.Similarly, adopt said method, can draw the predicated error of other each forecast models.

Step 103, according to described business factor historical data sequence and resource load amount historical data sequence, set up the correlation function of business factor and resource load amount.

Particularly, analysing and predicting system, first according to described business factor historical data sequence and resource load amount historical data sequence, adopt at least two function models to carry out the Function Fitting processing of described business factor and resource load amount, generate at least two candidate association functions; Then,, according to described business factor historical data sequence and resource load amount historical data sequence, calculate the matching residual error that adopts each candidate association function according to default least-squares calculation algorithm; Finally, the matching residual error of more each candidate association function, using the candidate association function of matching residual error minimum as described business factor and the correlation function of resource load amount.

Step 104, according to described correlation function, generate the expection resource load amount that described expection business factor is corresponding.

Particularly, analysing and predicting system, using described expection business factor as calculating parameter, according to described correlation function, generates the expection resource load amount that described expection business factor is corresponding then.

The embodiment of the present invention, by introducing business factor as influence factor in Forecasting Methodology, by the correlation function between matching business factor and resource load amount, has realized the Accurate Prediction of the resource load amount to-be to this application.The present embodiment has taken into full account the application deployment demand of different user and the load balancing characteristic of each application, for different user is disposed corresponding resource, dope accurately the following required resource load amount of different application, improve resource deployment and the maintenance mechanism of cloud computing platform, contributed to the utilization factor of the infrastructure resources that improves cloud computing system.

Further, in actual applications, be respectively applied in and in different time sections, there will be business factor to uprush or the situation of bust, for example, GPS application, during festivals or holidays, its business factor, registered user's number, simultaneously online user number or simultaneously online number of request can be apparently higher than working days.Therefore,, in order further to improve the prediction accuracy of the resource load amount Forecasting Methodology described in the embodiment of the present invention, need the prediction of resource load amount to be divided into resource load amount prediction during unusual day and the resource load amount prediction during normal day.Wherein, unusual day refers to repeatedly and occurs that in this day business factor and resource load amount are uprushed or the date of bust, and business factor and resource load amount are different from the period of normal day, and for example unusual day of GPS application is the legal festivals and holidays.Normal day is the date except unusual day, and for example, normal day of GPS application is the working day except the legal festivals and holidays.Particularly, the invention provides described resource load amount Forecasting Methodology embodiment bis-, as shown in Figure 3, the present embodiment two, except comprising above-mentioned steps 101～104, also comprised before described step 101:

Step 201, obtain business factor information.

Particularly, described analysing and predicting system obtains business factor information from database, and wherein, described business factor information at least comprises business factor and monitoring time.

Step 202, judge that monitoring time that described business factor information carries is whether within the default unusual day period, if, described business factor information is deposited in to unusual day historical data sequence of business factor, otherwise, deposit described business factor information in business factor normal day historical data sequence.

Particularly, described analysing and predicting system judges that monitoring time that business factor information carries is whether within the default unusual day period, if, described analysing and predicting system deposits described business factor information in business factor unusual day historical data sequence, otherwise described analysing and predicting system deposits described business factor information in business factor normal day historical data sequence.

Step 203, Gains resources charge capacity information.

Particularly, described analysing and predicting system is Gains resources charge capacity information from database, and wherein, described resource load amount information at least comprises business factor and monitoring time.

Step 204, judge that monitoring time that described resource load amount information carries is whether within the default unusual day period, if, described resource load amount information is deposited in to unusual day historical data sequence of resource load amount, otherwise, deposit described resource load amount information in resource load amount normal day historical data sequence.

Particularly, described analysing and predicting system judges that monitoring time that described resource load amount information carries is whether within the default unusual day period, if, described analysing and predicting system deposits described resource load amount information in resource load amount information unusual day historical data sequence, otherwise described analysing and predicting system deposits described resource load amount information in resource load amount information normal day historical data sequence.

Correspondingly, in above-described embodiment one step 101, the described business factor historical data sequence of obtaining respectively and resource load amount historical data sequence, be specially: described business factor historical data sequence is unusual day historical data sequence of business factor, described resource load amount historical data sequence is unusual day historical data sequence of resource load amount, or described business factor historical data sequence is the normal day historical data sequence of business factor, described resource load amount historical data sequence is the normal day historical data sequence of resource load amount.

Adopt respectively step 102～104 of above-described embodiment one, can dope respectively the expection resource load amount during unusual day and the expection resource load amount during normal day.

Again further, in resource load Forecasting Methodology, after described step 101, also comprise: respectively described business factor historical data sequence and resource load amount historical data sequence are carried out to data pre-service described in the various embodiments described above.

Particularly, analysing and predicting system carries out data pre-service to the business factor historical data sequence getting, and described analysing and predicting system carries out data pre-service to the resource load amount historical data sequence getting.Wherein, the pretreated object of data is: service operation supervisory system is in monitoring when described business factor, may occur the problems such as business factor that monitoring omits, monitors is abnormal.For avoiding the impact of these abnormal datas on subsequent step, further to improve the prediction accuracy of resource load amount Forecasting Methodology described in the present embodiment.Therefore, described analysing and predicting system needs respectively described business factor historical data sequence and described resource load amount historical data sequence to be carried out to data pre-service, can comprise following contents processing, for example, supplement the value lacking in historical data sequence, the Outlier Data in noise data, identification or deletion business factor historical data sequence and described resource load amount historical data sequence in smooth business factor historical data sequence and described resource load amount historical data sequence etc.Wherein, supplement the value lacking in historical data sequence, can, according to the value of front and back two adjacent datas, calculate on average supplementing of two adjacent datas, also can adopt interpolation calculation to supplement this missing values.

Further, in above-mentioned each resource load amount Forecasting Methodology embodiment, after described step 104, also comprise: described expection resource load amount is stored in database, so that service management system reads described expection resource load amount from described database, and generating resource deployment instruction according to described expection resource load amount, described resource management system is carried out resource deployment according to described resource deployment instruction.

Fig. 4 is the process flow diagram of charge capacity Forecasting Methodology embodiment tri-provided by the invention, and as shown in Figure 5, the method comprises:

Step 301, monitoring business factor.

Particularly, service operation supervisory system is according to Preset Time or predetermined period monitoring business factor.Wherein, wherein, described business factor can be specifically: registered user's number, while online user number, simultaneously online number of request etc.For example, GPS service application can be divided into three types, is respectively category-A GPS service application, category-B GPS service application and C class GPS service application.Wherein, category-A GPS service application, according to user scope, as number segment, is carried out relative static load balancing in advance, increases the granularity of division of number segment in the time of overload, and therefore category-A GPS service application Main Basis registered user number carries out load balancing.Category-B GPS service application is asked relative dynamic Resources allocation according to user, but owing to having incidence relation between such applied business request, the request of unique user can be bundled on specific application example within a certain period of time, therefore category-B GPS service application Main Basis simultaneously online user number carry out load balancing and resource deployment.C class GPS service application is distributed according to user's request dynamic, owing to there is no relevance between such application request, can will ask to be arbitrarily routed to resource node arbitrarily, therefore the simultaneously online number of request of C class GPS service application Main Basis is carried out the dynamic application of resource.

Step 302, the described business factor monitoring and monitoring time are stored in database, so that analysing and predicting system obtains business factor historical data sequence from described database, according to described business factor historical data sequence, generate expection business factor, according to described business factor historical data sequence and resource load amount historical data sequence, set up the correlation function of business factor and resource load amount, according to described correlation function, generate the expection resource load amount that described expection business factor is corresponding; Wherein, the monitoring time that in described business factor historical data sequence, each business factor information is carried is identical with the monitoring time that each resource load amount information in described resource load amount historical data sequence is carried respectively, described resource load amount historical data sequence is obtained from described database by described analysing and predicting system, and described resource load amount information is that the resource load amount monitoring and monitoring time are together stored to the information in described database by resource management system.

Particularly, when service operation supervisory system monitors business factor, described resource management system should be monitored the resource load amount that now business factor is corresponding simultaneously.Therefore, described service operation supervisory system, in the time monitoring described business factor, needs to record monitoring time simultaneously, and described business factor is stored in data together with monitoring time.Described resource management system, in the time monitoring described resource load amount, similarly needs to record monitoring time, and described resource load amount is stored in data together with monitoring time.Particularly, in order to guarantee business factor and the correspondence of resource load amount on monitoring time, described service operation supervisory system is monitored business factor according to Preset Time or predetermined period, and the business factor monitoring and monitoring time are together stored in database.Similarly, described resource management system is monitored resource load amount according to the Preset Time the same with described service operation supervisory system or predetermined period, and the resource load amount monitoring and monitoring time are together stored in database.

Further, in above-mentioned resource load amount Forecasting Methodology embodiment tri-, before described step 301, also comprise: receive the Monitoring instruction information that service management system sends, described Monitoring instruction information carries pre-monitoring business factor, so that described service operation supervisory system is according to described Monitoring instruction information, monitor described business factor.Particularly, when service management system receives the Request For Disposition application of user terminal transmission, service management system is by the load balancing feature for this application, select corresponding business factor, as registered user's number, simultaneously online user number or simultaneously online number of request, and send Monitoring instruction information to service operation supervisory system, wherein, in this Monitoring instruction information, carry business factor information.Described service operation supervisory system receives after described Monitoring instruction information, starts described business factor to monitor.

The embodiment of the present invention, by introducing business factor as influence factor in Forecasting Methodology, by the correlation function between matching business factor and resource load amount, has realized the Accurate Prediction of the resource load amount to-be to this application.The present embodiment has taken into full account the application deployment demand of different user and the load balancing characteristic of each application, for different user is disposed corresponding resource, dope accurately the following required resource load amount of different application, improve resource deployment and the maintenance mechanism of cloud computing platform, contributed to the utilization factor of the infrastructure resources that improves cloud computing system.

For the described resource load amount Forecasting Methodology that the embodiment of the present invention is provided is clearer, will apply as an example with GPS below.

The first, third party user (user of Request For Disposition application), to the application of service management system Request For Disposition, sends monitoring application request, and described monitoring application request carries pre-monitoring business factor, as while online user number.Described service management system, according to described monitoring application request, sends Monitoring instruction information to service operation supervisory system, and described Monitoring instruction information carries pre-monitoring business factor.Meanwhile, send monitoring resource command information to described resource management system.

The second, described service operation supervisory system is according to described Monitoring instruction information, online user number when monitoring GPS application.In this example, described service operation supervisory system by January, 2009～2012 that monitor GPS in year January application time online user number be stored in database.The monitoring periods of service operation supervisory system is sky, totally 1095 Monitoring Data, as shown in Figure 5.Meanwhile, described resource management system is according to described monitoring resource command information, and monitoring GPS is applied in the resource load amount of every day between in January, 2009～2012 year January, and the resource load amount monitoring is stored in database.The monitoring periods of resource management system is sky, totally 1095 Monitoring Data, as shown in Figure 6.

The 3rd, analysing and predicting system is according to the resource load amount of GPS on January 1,1 day～2012 January in 2009 application time online user number and resource load amount predict future half a year.In in January, 2009～2012 year 1 month, unusual day period information comprises: the legal festivals and holidays.Other except the legal festivals and holidays are normal day.Specifically can from Fig. 5 and Fig. 6, find out, catastrophe point is the resource load amount that unusual day period monitored GPS application while online user number and GPS application.To the prediction of resource load amount half a year in future, be implemented as follows:

First, analysing and predicting system obtains the business factor information of monitoring on January 1,1 day～2012 January in 2009 from database, and judge that respectively monitoring time that these 1095 business factor information carry is whether within above-mentioned unusual day period, if, described business factor information is deposited in to unusual day historical data sequence of business factor, otherwise, deposit described business factor information in business factor normal day historical data sequence.In like manner analysing and predicting system obtains the resource load amount information of monitoring on January 1,1 day～2012 January in 2009 from database, and judge that respectively monitoring time that these 1095 resource load amounts carry is whether within above-mentioned unusual day period, if, described resource load amount information is deposited in to unusual day historical data sequence of resource load amount, otherwise, deposit described resource load amount information in resource load amount normal day historical data sequence.Described analysing and predicting system is first applied in following half a year of normal resource load amount in a few days according to the normal day historical data sequence of business factor and a normal day historical data sequence prediction GPS of resource load amount, then is applied in the resource load amount in following unusual day of half a year according to unusual day historical data sequence of business factor and unusual day historical data sequence prediction GPS of resource load amount.

Then, described analysing and predicting system obtains respectively business factor historical data sequence and resource load amount historical data sequence, according to described business factor historical data sequence, generates expection business factor.

Wherein, described business factor historical data sequence comprises the normal day historical data sequence of business factor and unusual day historical data sequence of business factor.Particularly, a normal business factor day historical data sequence is divided into two arrays by monitoring time, in order to lower computation complexity, the business factor information on Dec 31st, 1 day 1 January in 2009 in normal business factor day historical data sequence is divided into the first array by this example, and the business factor information on January 1st, 2012 is divided into the second array.Adopting three kinds of forecast models is phase space reconfiguration forecast model, grey forecasting model, three Smoothing Prediction models, respectively based on described the first array, dope the business factor on January 1st, 2012, difference between the business factor in 1 day January in 2012 in business factor and second array in the 1 day January in 2012 that comparison prediction goes out, calculates predicated error that there emerged a forecast model according to described difference.This predicated error can be characterized by number percent number, for example, and the number percent of the business factor in 1 day January in 2012 in difference and the second array.Concrete result of calculation is as shown in table 1.

The predicated error table of comparisons of three kinds of forecast models of table 1

As can be seen from Table 1, the predicated error minimum of phase space reconfiguration forecast model.Therefore, this example adopts phase space reconfiguration forecast model, draws expection business factor according to the normal day historical data sequence of described business factor.As shown in Figure 7, according to the normal day historical data sequence of the business factor on January 1st, 1 day 1 January in 2009, draw the normal day business factor of expection on July 1st, 2 days 1 January in 2012.In like manner, according to unusual day historical data sequence of the business factor on January 1st, 1 day 1 January in 2009, draw the unusual day business factor of expection on July 1st, 2 days 1 January in 2012.As shown in Figure 7, unusual day business factor of normal day business factor of described expection and described expection formed the expection business factor on July 1st, 2 days 1 January in 2012.

Wherein, phase space reconfiguration forecast model is realized principle and is, first determines phase space parameter: in phase space reconfiguration process, have two very important parameters: delay time T and embedding dimension m.Their selections are directly connected to the quality of phase space reconfiguration.Utilize C-C method to adopt correlation integral determine delay time T and embed dimension m, computing formula is as follows:

$C(m,N,r,t)=\frac{2}{M(M-1)}\underset{1\&le;i<j\&le;M}{\mathrm{\&Sigma;}}H(r-||x_i-x_j|\left|\right);r>0---\left(1\right)$

In above formula (1) $H\left(r\right)=\left\{\begin{array}{c}1,r\&GreaterEqual;0\\ 0,r<0\end{array}\right.---\left(2\right)$

Can be found out by above-mentioned formula (1), correlation integral is cumulative distribution function, represents in phase space that distance between any two points is less than the probability of r.Here distance between points represents with the Infinite Norm of the difference of vector.Definition test statistics: S (m, N, r, t)=C (m, N, r, t)-C ^m(1, N, r, t) describes the correlativity of Nonlinear Time Series, and found delay time T and embedded dimension m by statistic S (m, N, r, t).

Wherein, the computation process of statistic S (m, N, r, t) is: Time Series is become to t mutual nonoverlapping subsequence, and t is reconstruct time delay, that is:

x ¹={x(k),k=1,1+t,...,1+N-t}

x ²={x(k),k=2,2+t,...,2+N-t}

...

x ^t={x(k),k=t,2t,...,N} （4）

In above-mentioned (4) formula, the integral multiple that N is t, statistic S defined above (m, N, r, t) adopts piecemeal average strategy, as shown in the formula:

$S(m,N,r,t)=\frac{1}{t}\underset{s=1}{\overset{t}{\mathrm{\&Sigma;}}}[C_s(m,\frac{N}{t},r,t)-C_s^m(1,\frac{N}{t},r,t)]---\left(5\right)$

When in (5) formula when N → ∞,

$S(m,r,t)=\frac{1}{t}\underset{s=1}{\overset{t}{\mathrm{\&Sigma;}}}[C_s(m,r,t)-C_s^m(1,r,t)]---\left(6\right)$

Optimum delay t can get the time point of S (m, r, t) to the mutual difference minimum of all radius r, selects two minimum and maximum radius r, definition residual quantity:

ΔS(m,t)=max{S(m,r _j,t)}-min{S(m,r _j,t)} （7）

Δ S (m, t) has measured the maximum deviation of S (m, r, t) pair radius r.Because Δ S (m, t) is always positive number, optimum delay τ can get the corresponding time point of first local minimum of Δ S (m, t)～t.Due to

all reflect former seasonal effect in time series autocorrelation performance, definition index:

$S_{\mathrm{cor}}\left(t\right)=\mathrm{\&Delta;}\stackrel{\&OverBar;}{S}\left(t\right)+\left|\stackrel{\&OverBar;}{S}\left(t\right)\right|---\left(8\right)$

Find S _cor(t) the corresponding t of global minimum can obtain the best window t that embeds _w.The embedding window method of phase space reconfiguration thinks that choosing of delay time T should not be independent of embedding dimension m, embeds window t and should depend on _w=(m-1) τ, can calculate thus and embed dimension m.

Then, carry out phase space reconfiguration prediction according to the above-mentioned delay time T calculating and embedding dimension m.Can be divided into global approach and two kinds of methods of local method according to the mode of attractor in matching phase space.So-called global approach is whole in matching object using in track, finds out its rule, obtain f ( ^*), the trend of prediction locus thus.On this theoretical method, be feasible, but in the time of trajectory of phase space more complicated, be difficult to make prediction accurately.Local method is using the last point of trajectory of phase space as central point, decentering is put to nearest some tracing points as reference point, then these reference points are made to matching, then estimate under track the trend of a bit, finally from the coordinate of the tracing point that dopes, isolate required predicted value.

In local method, according to the value of the neighbor point of central point or walk always to predict trajectory of phase space.Such as, we will predict the weather of tomorrow, can find in history and the most close one day of the natural situation of today the weather forecasting value using the weather conditions of second day of that day as tomorrow.Take the local method of first approximation matching as example, a small neighbourhood of ordering with n in phase space infer under any tendency.So-called first approximation refers to X (t+1)=a+bX (t) carrys out matching n point small neighbourhood around.

In the process of phase space reconfiguration, sequence length between X (t+1)=a+bX (t) when establishing N and being, M is the number of phase space mid point, M=N-(m-1) * τ, the expression formula of trajectory of phase space is:

X(t+τ)=f(X(t)) (9)

Wherein, X (t+ τ) can be considered the mapping of f (X (t)),

X(t)=[x(t),x(t+τ),...,x(t+(m-1)τ)] (10)

Mapping in above-mentioned (10) formula can be expressed as following time series:

X(1)=[x(1),x(1+τ),...,x(1+(m-1)τ)]

X(2)=[x(2),x(2+τ),...,x(2+(m-1)τ)]

...

X(M)=[x(M),x(M+τ),...,x(N)] (11)

If central point X _mreference vector collection { X _mi, i=1,2 ..., the phase point set after its evolution of q k step is { X _mi+k, first order local area linear fit is as minor function:

X _Mi+k＝a _ke+b _kX _Mi,i＝1,2,...,q (12)

Can obtain according to weighted least-squares method:

$\underset{i=1}{\overset{q}{\mathrm{\&Sigma;}}}{P}_i\left[\underset{j=1}{\overset{m}{\mathrm{\&Sigma;}}}{(x_{\mathrm{Mi}+k}^j-a_k-b_k{x}_{\mathrm{Mi}}^j)}^2\right]=\min ---\left(13\right)$

Wherein,

reference vector X _mij element.Above formula (13) is regarded as about unknown number a _kand b _kbinary function, ask local derviation abbreviation to draw on above-mentioned formula (13) both sides:

$\left\{\begin{array}{c}{a}_k\underset{i=1}{\overset{q}{\mathrm{\&Sigma;}}}{P}_i\underset{j=1}{\overset{m}{\mathrm{\&Sigma;}}}{x}_{\mathrm{Mi}}^j+b_k\underset{i=1}{\overset{q}{\mathrm{\&Sigma;}}}{P}_i\underset{j=1}{\overset{m}{\mathrm{\&Sigma;}}}{\left(x_{\mathrm{Mi}}^j\right)}^2=\underset{i=1}{\overset{q}{\mathrm{\&Sigma;}}}{P}_i\underset{j=1}{\overset{m}{\mathrm{\&Sigma;}}}{x}_{\mathrm{Mi}+k}^j{x}_{\mathrm{Mi}}^k\\ a_{k}m+b_k\underset{i=1}{\overset{q}{\mathrm{\&Sigma;}}}{P}_i\underset{j=1}{\overset{m}{\mathrm{\&Sigma;}}}{x}_{\mathrm{Mi}}^j=\underset{i=1}{\overset{q}{\mathrm{\&Sigma;}}}{P}_i\underset{j=1}{\overset{m}{\mathrm{\&Sigma;}}}{x}_{\mathrm{Mi}+k}^j\end{array}\right.---\left(14\right)$

Above-mentioned (14) formula is rewritten into matrix form is:

$\left(\begin{array}{cc}\mathrm{\&alpha;}& \mathrm{\&beta;}\\ m& \mathrm{\&alpha;}\end{array}\right)\left(\begin{array}{c}{a}_k\\ b_k\end{array}\right)=\left(\begin{array}{c}{e}_k\\ f_k\end{array}\right)---\left(15\right)$

Wherein: $\mathrm{\&alpha;}=\underset{i=1}{\overset{q}{\mathrm{\&Sigma;}}}{P}_i\underset{j=1}{\overset{m}{\mathrm{\&Sigma;}}}{x}_{\mathrm{Mi}}^j,$ $\mathrm{\&beta;}=\underset{i=1}{\overset{q}{\mathrm{\&Sigma;}}}{P}_i\underset{j=1}{\overset{m}{\mathrm{\&Sigma;}}}{\left(x_{\mathrm{Mi}}^j\right)}^2,$ $e_k=\underset{i=1}{\overset{q}{\mathrm{\&Sigma;}}}{P}_i\underset{j=1}{\overset{m}{\mathrm{\&Sigma;}}}{x}_{\mathrm{Mi}+k}^j{x}_{\mathrm{Mi}}^j,$ $f_k=\underset{i=1}{\overset{q}{\mathrm{\&Sigma;}}}{P}_i\underset{j=1}{\overset{m}{\mathrm{\&Sigma;}}}{x}_{\mathrm{Mi}+k}^j$ :

$\left(\begin{array}{c}{a}_k\\ b_k\end{array}\right)={\left(\begin{array}{cc}\mathrm{\&alpha;}& \mathrm{\&beta;}\\ m& \mathrm{\&alpha;}\end{array}\right)}^{-1}\left(\begin{array}{c}{e}_k\\ f_k\end{array}\right)---\left(16\right)$

According to a trying to achieve _k, b _k, substitution k step predictor formula X _m+1=a _ke+b _kx _m, can obtain the phase point predicted value developing after k step:

X _M+k＝(x _M+k,x _M+k+τ，...,x _M+k+(m-1)τ) （17）

Here X, _m+kin m element x _{m+k+ (m-1) τ}be the k step predicted value of former sequence.

Based on above-mentioned phase space reconfiguration prediction algorithm, through calculating

in the time of t=12, obtain first local minizing point, therefore get τ=12 as optimum delay.S _cor(t) in the time of t=52, obtain overall smallest point, i.e. t _w=52.According to t _w=(m-1) τ, therefore gets m=5, carries out phase space reconfiguration according to formula (11).

The principle that realizes of grey forecasting model is, GM (1,1) model is a kind of dynamic sequence disposal route in gray system theory, and it is only to comprise univariate differential equation of first order.

If x ⁽⁰⁾for nonnegative sequence:

x ⁽⁰⁾={x ⁽⁰⁾(1),x ⁽⁰⁾(2),...,x ⁽⁰⁾(n)}（18）

X ⁽¹⁾for x ⁽⁰⁾the cumulative sequence of single order:

x ⁽¹⁾={x ⁽¹⁾(1),x ⁽¹⁾(2),...,x ⁽¹⁾(n)} （19）

Wherein: x ⁽⁰⁾(k) >0, k=1,2 ..., k=1,2 ..., n, the corresponding differential equation of GM (1,1) model is:

$\frac{{\mathrm{dx}}^{\left(1\right)}\left(t\right)}{\mathrm{dt}}+\mathrm{\&alpha;}{x}^{\left(1\right)}\left(t\right)=u,t=\mathrm{1,2},...,n---\left(20\right)$

In formula (20), α is the grey number of development; U is grey action.

Parameter vector

utilize least square method to solve:

wherein,

$B=\left|\begin{array}{cc}-\frac{1}{2}(x^{\left(1\right)}\left(1\right)+x^{\left(1\right)}\left(2\right))& 1\\ -\frac{1}{2}(x^{\left(1\right)}\left(2\right)+x^{\left(1\right)}\left(3\right))& 1\\ ...& ...\\ -\frac{1}{2}(x^{\left(1\right)}(n-1)+x^{\left(1\right)}\left(n\right))& 1\end{array}\right|,$ $Y_n=\left|\begin{array}{c}{x}^{\left(0\right)}\left(2\right)\\ x^{\left(0\right)}\left(3\right)\\ ...\\ x^{\left(0\right)}\left(n\right)\end{array}\right|$

Parameter alpha, u substitution formula (20) that calculating is tried to achieve, and solve, x got ⁽¹⁾(0)=x ⁽⁰⁾(1), obtain grey forecasting model:

${\hat{x}}^{\left(1\right)}(t+1)=[x^{\left(0\right)}\left(1\right)-\frac{u}{\mathrm{\&alpha;}}]e^{-\mathrm{\&alpha;t}}+\frac{u}{\mathrm{\&alpha;}}---\left(21\right)$

What obtained by formula (21) is accumulated value successively

the analogue value, then obtain real predicted value by a regressive:

${\hat{x}}^{\left(0\right)}(t+1)={\hat{x}}^{\left(1\right)}(t+1)-{\hat{x}}^{\left(1\right)}\left(t\right)---\left(22\right)$ Wherein, t=1,2 ..., n.

Based on the principle of above-mentioned grey forecasting model, first utilize least square method to solve parameter vector:

$\widehat{\mathrm{\&alpha;}}={\left(B^{T}B\right)}^{-1}{B}^T{Y}_n=\left(\begin{array}{c}0.0347\\ 52.8371\end{array}\right)$

According to $\widehat{\mathrm{\&alpha;}}=\left(\begin{array}{c}a\\ u\end{array}\right)=\left(\begin{array}{c}0.0347\\ 52.8371\end{array}\right)$ Known a=0.0347, u=52.8371.By solving in parameter a, the u substitution differential equation, can obtaining gray prediction response function be:

${\hat{x}}^{\left(1\right)}(t+1)=[x^{\left(0\right)}\left(1\right)-\frac{u}{\mathrm{\&alpha;}}]e^{-\mathrm{\&alpha;t}}+\frac{u}{\mathrm{\&alpha;}}=-{1439.8629e}^{-0.0347t}+1522.6809$

Wherein, the poor ratio of posteriority is C=32.83%, and the little probability of error is P=91.64%, and known precision of prediction grade is better.

The principle that realizes of three Smoothing Prediction models is that Smoothing Prediction principle is to utilize the impact of historical data smoothly being eliminated to enchancement factor.In the time that having shaped form tendency, historical data sequence need to use three times exponential smoothing.The ultimate principle of three Smoothing Prediction models be to raw data through three exponential smoothing process after, in order to estimate quadratic polynomial parameter, thereby set up forecast model.As follows:

If time series is X ₁, X ₂, X ₃..., X _n, representing exponential smoothing value with S, t phase single exponential smoothing value is designated as

double smoothing value is designated as

three times exponential smoothing value is designated as

wherein, the following formula of definite employing of level and smooth initial value:

$S_0^{\left(1\right)}=S_0^{\left(2\right)}=S_0^{\left(3\right)}=\frac{X_1+{\mathrm{<figure-callout\; id="2"\; label="X"\; filenames="HDA00002274974500062.png,HDA00002274974500071.png"\; state="\{\{state\}\}">X</figure-callout>}}_2+{\mathrm{<figure-callout\; id="3"\; label="X"\; filenames="HDA00002274974500062.png,HDA00002274974500071.png"\; state="\{\{state\}\}">X</figure-callout>}}_3}{3}---\left(23\right)$

Exponential smoothing value computing formula is:

$S_t^{\left(1\right)}={\mathrm{\&alpha;}X}_t+(1-\mathrm{\&alpha;})S_{t-1}^{\left(1\right)}---\left(24\right)$

$S_t^{\left(2\right)}=\mathrm{\&alpha;}{S}_t^{\left(1\right)}+(1-\mathrm{\&alpha;})S_{t-1}^{\left(2\right)}---\left(25\right)$

$S_t^{\left(3\right)}=\mathrm{\&alpha;}{S}_t^{\left(2\right)}+(1-\mathrm{\&alpha;})S_{t-1}^{\left(3\right)}---\left(26\right)$

Wherein, α ∈ [0,1] is smoothing factor, smoothing factor α can determine its fair-sized by actual conditions, generally chooses according to Minimum Mean Square Error, respectively different α values is carried out to Smoothing Prediction, calculate respectively mean square deviation, get the α value of Minimum Mean Square Error as smoothing factor.Be the index prediction value Y that T days, radix are t days to predetermined period _t+T, the mathematical model of its three exponential smoothings is:

Y _t+T＝a _t+b _tT+c _tT ² （27）

Wherein: a _t, b _t, c _tbe smoothing factor, computing formula is:

$a_t={{3S}_t}^{\left(1\right)}-{{3S}_t}^{\left(2\right)}+{{3S}_t}^{\left(3\right)}---\left(28\right)$

$b_t=\frac{\mathrm{\&alpha;}}{2{(1-\mathrm{\&alpha;})}^2}[(6-5\mathrm{\&alpha;})S_t^{\left(1\right)}-2(5-4\mathrm{\&alpha;})S_t^{\left(2\right)}+(4-3\mathrm{\&alpha;})S_t^{\left(3\right)}]---\left(29\right)$

$c_t=\frac{{\mathrm{\&alpha;}}^2}{2{(1-\mathrm{\&alpha;})}^2}[{S_t}^{\left(1\right)}-{{2S}_t}^{\left(2\right)}+{S_t}^{\left(3\right)}]$

In three exponential smoothing forecasting processes, first determine level and smooth initial value:

$S_0^{\left(1\right)}=S_0^{\left(2\right)}=S_0^{\left(3\right)}=\frac{X_1+{\mathrm{<figure-callout\; id="2"\; label="X"\; filenames="HDA00002274974500062.png,HDA00002274974500071.png"\; state="\{\{state\}\}">X</figure-callout>}}_2+{\mathrm{<figure-callout\; id="3"\; label="X"\; filenames="HDA00002274974500062.png,HDA00002274974500071.png"\; state="\{\{state\}\}">X</figure-callout>}}_3}{3}=1541,$

Secondly respectively to carrying out Smoothing Prediction take 0.02 as the different α values at interval between [0,1], get α=0.72 value of Minimum Mean Square Error as smoothing factor.If predetermined period T=1,2 ..., 5 days, the index prediction value Y that radix is t=1095 days _1095+T, utilization index smooth value computing formula draws respectively substitution smoothing factor computing formula a ₁₀₉₅=29.3925, b ₁₀₉₅=-0.4814, c ₁₀₉₅=0.0212, three exponential smoothing computing formula Y of while online user number _1095+T=29.3925-0.4814T+0.0212T ².

Subsequently, described analysing and predicting system, according to the normal day historical data sequence of the described business factor getting and the normal day historical data sequence of resource load amount, is set up the correlation function of business factor and resource load amount.

Particularly, from historical data Fig. 6, can observe and draw, along with the growth of while online user number, resource load amount also presents direct proportion rising tendency.Therefore, analysing and predicting system can utilize respectively four kinds of default growth form functions to carry out Function Fitting processing to resource load amount and while online user number.These four kinds of growth form functions comprise: linear function type, polynomial function type, exponential function type and power function type.

Linear function: y=ax+b;

Polynomial function: y=a ₁x ⁿ+ a ₂x ^n-1+ a ₃x ^n-2+ ... .+a _n-1x+a _n;

Exponential function: y=ae ^bx;

Power function: y=ax ^b.

Wherein, in polynomial function, the selection of n value should not be too high, and n value is higher, and the Fitting Calculation process is just more complicated.

In fact, Function Fitting processing procedure is exactly, the business factor and the resource load amount that in normal business factor day historical data sequence and the normal day historical data sequence of resource load amount, carry identical monitoring time are updated to respectively in function, solve in above-mentioned function, as coefficient a and b in linear function, coefficient a1, a2 in polynomial function, a3 ... an, the coefficient a in exponential function and power function and power exponent b.In above-mentioned function, after having solved of coefficient or index, four candidate association functions are drawn.Subsequently, according to the normal day historical data sequence of described business factor and the normal day historical data sequence of resource load amount, calculate the matching residual error that adopts each candidate association function according to default least-squares calculation algorithm, to select the candidate association function of matching residual error minimum as the correlation function of while online user number and resource load amount.

Wherein, described least-squares calculation is calculated ratio juris and is, according to the business factor and the resource load amount that carry identical monitoring time in the normal day historical data sequence of business factor and the normal day historical data sequence of resource load amount, can obtain n node, as (x ₁, y ₁), (x ₂, y ₂), (x ₃, y ₃) ..., (x _n, y _n).As shown in Figure 8, candidate association function is at identical x ₁, x ₂, x ₃..., x _nthe y that coordinate place is corresponding ₁', y ₂', y ₃' ..., y _n' value respectively with y ₁, y ₂, y ₃..., y _nthe quadratic sum of difference, be matching residual error.Specifically be expressed as:

adopt four kinds of growth form functions to carry out the matching residual error of Function Fitting candidate association function after treatment and each candidate association function, as shown in table 2.

The matching residual error table of comparisons of table 2, candidate association function and each candidate association function

From above-mentioned table 2, the matching residual error minimum of linear function y=0.04x-45.00, this linear function y=0.04x-45.00 is the correlation function of GPS application while online user number and resource load amount.

Finally, according to described linear function y=0.04x-45.00, using the above-mentioned expection business factor doping as calculating parameter, i.e. x in this linear function, calculates expection resource load amount y.According to the expection business factor on the July 1st, 2 days 1 January in 2012 described in Fig. 8, calculate the expection resource load amount on July 1,2 days～2012 January in 2012 as shown in Figure 9.

One of ordinary skill in the art will appreciate that: all or part of step that realizes above-mentioned each embodiment of the method can complete by the relevant hardware of programmed instruction.Aforesaid program can be stored in a computer read/write memory medium.This program, in the time carrying out, is carried out the step that comprises above-mentioned each embodiment of the method; And aforesaid storage medium comprises: various media that can be program code stored such as ROM, RAM, magnetic disc or CDs.

As shown in figure 10, the structural representation of analysing and predicting system embodiment mono-provided by the invention, as shown in FIG., described analysing and predicting system comprises: the first acquisition module 1, the first processing module 2, set up module 3 and the second processing module 4.Wherein, described the first acquisition module 1 is for obtaining respectively business factor historical data sequence and resource load amount historical data sequence, wherein, the monitoring time that in described business factor historical data sequence, each business factor information is carried is identical with the monitoring time that each resource load amount information in described resource load amount historical data sequence is carried respectively, described business factor information is that the business factor monitoring and monitoring time are together stored to the information in database by service operation supervisory system, described resource load amount information is that the resource load amount monitoring and monitoring time are together stored to the information in described database by resource management system.Described the first processing module 2, for according to described business factor historical data sequence, generates expection business factor.The described module 3 of setting up, for according to described business factor historical data sequence and resource load amount historical data sequence, is set up the correlation function of business factor and resource load amount.Described the second processing module 4, for according to described correlation function, generates the expection resource load amount that described expection business factor is corresponding.

Described in the embodiment of the present invention, analysing and predicting system, by introducing business factor as influence factor in forecasting process, by the correlation function between matching business factor and resource load amount, has been realized the Accurate Prediction of the resource load amount to-be to this application.The present embodiment has taken into full account the application deployment demand of different user and the load balancing characteristic of each application, for different user is disposed corresponding resource, dope accurately the following required resource load amount of different application, improve resource deployment and the maintenance mechanism of cloud computing platform, contributed to the utilization factor of the infrastructure resources that improves cloud computing system.

Further, as shown in figure 11, above-mentioned analysing and predicting system embodiment also comprises: the second acquisition module 5, judge module 6 and execution module 7.Wherein, described the second acquisition module 5 is for obtaining business factor information.Described judge module 6 whether within the default unusual day period, if so, generates the first judgement information for the monitoring time that judges described business factor information and carry, otherwise, generate the second judgement information.Described execution module, for according to described the first judgement information, deposits described business factor information in business factor unusual day historical data sequence; According to described the second judgement information, deposit described business factor information in business factor normal day historical data sequence.Described the second acquisition module 7 is also for Gains resources charge capacity information.Described judge module is also for judging that monitoring time that described resource load amount information carries whether within the default unusual day period, if so, generates the 3rd judgement information, otherwise, generate the 4th judgement information.Described execution module also, for according to described the 3rd judgement information, deposits described resource load amount information in resource load amount unusual day historical data sequence; According to described the second judgement information, deposit described resource load amount information in resource load amount normal day historical data sequence.Correspondingly, the described business factor historical data sequence that described the first acquiring unit 1 obtains is unusual day historical data sequence of business factor, described resource load amount historical data sequence is unusual day historical data sequence of resource load amount, or, the described business factor historical data sequence that described the first acquiring unit obtains is the normal day historical data sequence of business factor, and described resource load amount historical data sequence is the normal day historical data sequence of resource load amount.

Again further, above-mentioned analysing and predicting system embodiment also comprises: data preprocessing module.Wherein, described data preprocessing module is for carrying out data pre-service to described business factor historical data sequence and resource load amount historical data sequence respectively.

Further, above-mentioned analysing and predicting system embodiment also comprises: memory module.Described memory module is for being stored to database by described expection resource load amount, so that described service management system reads described expection resource load amount from described database, and generating resource deployment instruction according to described expection resource load amount, described resource management system is carried out resource deployment according to described resource deployment instruction.

Wherein, in above-mentioned analysing and predicting system embodiment, described the first processing module can adopt structure as shown in figure 12 to realize.Described the first processing module 2 comprises particularly: the first processing unit 21 and the second processing unit 22.Wherein, described the first processing unit 21, for according to described business factor historical data sequence, draws the predicated error of at least two kinds of default forecast models.Described the second processing unit 22, for according to described business factor historical data sequence, adopts the forecast model of described predicated error minimum to calculate expection business factor.The described module 3 of setting up can adopt structure as shown in figure 13 to realize, and particularly, the described module 3 of setting up comprises: matching unit 31, the 3rd processing unit 32 and comparing unit 33.Wherein, described matching unit 31 is for according to described business factor historical data sequence and resource load amount historical data sequence, adopt at least two function models to carry out the Function Fitting processing of described business factor and resource load amount, generate at least two candidate association functions.Described the 3rd processing unit 32, for according to described business factor historical data sequence and resource load amount historical data sequence, calculates the matching residual error that adopts each candidate association function according to default least-squares calculation algorithm.Described comparing unit 33 is for the matching residual error of more each candidate association function, using the candidate association function of matching residual error minimum as described business factor and the correlation function of resource load amount.

As shown in figure 14, the structural representation of service operation supervisory system embodiment mono-provided by the invention.As shown in FIG., described in the present embodiment, service operation supervisory system comprises: monitoring modular 8 and memory module 9.Wherein, described monitoring modular 8 is for monitoring business factor.Described memory module 9 is for being stored to database by the described business factor and the monitoring time that monitor, so that analysing and predicting system obtains business factor historical data sequence from described database, according to described business factor historical data sequence, generate expection business factor, according to described business factor historical data sequence and resource load amount historical data sequence, set up the correlation function of business factor and resource load amount, according to described correlation function, generate the expection resource load amount that described expection business factor is corresponding.Wherein, the monitoring time that in described business factor historical data sequence, each business factor information is carried is identical with the monitoring time that each resource load amount information in described resource load amount historical data sequence is carried respectively, described resource load amount historical data sequence is obtained from described database by described analysing and predicting system, and described resource load amount information is that the resource load amount monitoring and monitoring time are together stored to the information in described database by resource management system.

Described in the embodiment of the present invention, service operation supervisory system is by monitoring business factor, can make analysing and predicting system in forecasting process, introduce business factor as influence factor, by the correlation function between matching business factor and resource load amount, realize the Accurate Prediction of the resource load amount to-be to this application.The present embodiment has taken into full account the application deployment demand of different user and the load balancing characteristic of each application, for different user is disposed corresponding resource, dope accurately the following required resource load amount of different application, improve resource deployment and the maintenance mechanism of cloud computing platform, contributed to the utilization factor of the infrastructure resources that improves cloud computing system.

Further, above-mentioned service operation supervisory system embodiment also comprises receiver module.The Monitoring instruction information that described receiver module sends for receiving service management system, described Monitoring instruction information carries pre-monitoring business factor, so that described service operation supervisory system is according to described Monitoring instruction information, monitors described business factor.

As shown in figure 15, the structural representation of cloud computing system framework embodiment mono-provided by the invention.As shown in Figure 15, described in the present embodiment, cloud computing system framework comprises: service management system 11, database 12, resource management system 13, analysing and predicting system 14 and service operation supervisory system 15.Wherein, described analysing and predicting system 14 adopts the analysing and predicting system described in the various embodiments described above.Described service operation supervisory system 15 adopts the service operation supervisory system described in the various embodiments described above.The specific implementation principle of described analysing and predicting system and described service operation supervisory system, referring to disclosed related content in above-mentioned corresponding embodiment, repeats no more herein.Described service management system 11 is for realizing the functions such as sla management, service catalogue management, accounting management, customer account management.Wherein sla management function can, to SLA template and correlative factor thereof, as ALM etc., be checked.Service catalogue management function can be inquired about service, be opened, closes and deletion action.Accounting management function can increase charging policy, deletes and order is inquired about, renewed a contract and settles accounts.Customer account management function can inquire about and revise client-related information.Resource management system 13 is for realizing the functions such as monitoring resource, resource control, dynamic-configuration, dynamic migration.Wherein monitoring resource function is carried out Real-Time Monitoring mainly for CPU usage, internal memory use amount, hard disk use amount, the network bandwidth occupancy of physical machine and virtual machine.Resource control function is mainly opened, closes, is restarted, time-out, recovery, dormancy, the life cycle control such as wakes up virtual machine, and physical machine such as is opened, restarts at the life cycle control.Dynamic-configuration function is carried out the configuration of CPU or internal memory in the situation that moving mainly for virtual machine.Dynamic migration function is mainly for virtual machine in the situation that moving, and virtual machine moves to another host from a host quick and smooth.

In several embodiment provided by the present invention, should be understood that, disclosed system, apparatus and method, can realize by another way.For example, device embodiment described above is only schematic, for example, the division of described unit, be only that a kind of logic function is divided, when actual realization, can have other dividing mode, for example multiple unit or assembly can in conjunction with or can be integrated into another system, or some features can ignore, or do not carry out.

The described unit as separating component explanation can or can not be also physically to separate, and the parts that show as unit can be or can not be also physical locations, can be positioned at a place, or also can be distributed in multiple network element.Can select according to the actual needs some or all of unit wherein to realize the object of the present embodiment scheme.

In addition, the each functional unit in each embodiment of the present invention can be integrated in a processing unit, can be also that the independent physics of unit exists, and also can be integrated in a unit two or more unit.Above-mentioned integrated unit both can adopt the form of hardware to realize, and the form that also can adopt hardware to add SFU software functional unit realizes.

The integrated unit that the above-mentioned form with SFU software functional unit realizes, can be stored in a computer read/write memory medium.Above-mentioned SFU software functional unit is stored in a storage medium, comprise that some instructions (can be personal computers in order to make a computer equipment, server, or the network equipment etc.) or processor (processor) carry out the part steps of method described in each embodiment of the present invention.And aforesaid storage medium comprises: USB flash disk, portable hard drive, ROM (read-only memory) (Read-Only Memory, be called for short ROM), the various media that can be program code stored such as random access memory (Random Access Memory, be called for short RAM), magnetic disc or CD.

It should be noted that: for aforesaid each embodiment of the method, for simple description, therefore it is all expressed as to a series of combination of actions, but those skilled in the art should know, the present invention is not subject to the restriction of described sequence of movement, because according to the present invention, some step can adopt other orders or carry out simultaneously.Secondly, those skilled in the art also should know, the embodiment described in instructions all belongs to preferred embodiment, and related action and module might not be that the present invention is necessary.

In the above-described embodiments, the description of each embodiment is all emphasized particularly on different fields, in certain embodiment, there is no the part of detailed description, can be referring to the associated description of other embodiment.Those skilled in the art can be well understood to, for convenience and simplicity of description, the system of foregoing description, the specific works process of device and unit, can, with reference to the corresponding process in preceding method embodiment, not repeat them here.

Finally it should be noted that: above embodiment only, in order to technical scheme of the present invention to be described, is not intended to limit; Although the present invention is had been described in detail with reference to previous embodiment, those of ordinary skill in the art is to be understood that: its technical scheme that still can record aforementioned each embodiment is modified, or part technical characterictic is wherein equal to replacement; And these modifications or replacement do not make the essence of appropriate technical solution depart from the spirit and scope of various embodiments of the present invention technical scheme.

Claims (10)

1. a resource load amount Forecasting Methodology, is characterized in that, comprising:

Obtain respectively business factor historical data sequence and resource load amount historical data sequence, wherein, the monitoring time that in described business factor historical data sequence, each business factor information is carried is identical with the monitoring time that each resource load amount information in described resource load amount historical data sequence is carried respectively;

According to described business factor historical data sequence, generate expection business factor;

According to described business factor historical data sequence and resource load amount historical data sequence, set up the correlation function of business factor and resource load amount;

According to described correlation function, generate the expection resource load amount that described expection business factor is corresponding.

2. resource load amount Forecasting Methodology according to claim 1, is characterized in that, described obtain respectively business factor historical data sequence and resource load amount historical data sequence before, also comprise:

Obtain business factor information;

Judge that monitoring time that described business factor information carries is whether within the default unusual day period, if, described business factor information is deposited in to unusual day historical data sequence of business factor, otherwise, deposit described business factor information in business factor normal day historical data sequence;

Gains resources charge capacity information;

Judge that monitoring time that described resource load amount information carries is whether within the default unusual day period, if, described resource load amount information is deposited in to unusual day historical data sequence of resource load amount, otherwise, deposit described resource load amount information in resource load amount normal day historical data sequence;

Described obtaining respectively in business factor historical data sequence and resource load amount historical data sequence, described business factor historical data sequence is unusual day historical data sequence of business factor, described resource load amount historical data sequence is unusual day historical data sequence of resource load amount, or described business factor historical data sequence is the normal day historical data sequence of business factor, described resource load amount historical data sequence is the normal day historical data sequence of resource load amount.

3. resource load amount Forecasting Methodology according to claim 1 and 2, is characterized in that, described according to described correlation function, after calculating the described expection resource load amount that described expection business factor is corresponding, also comprises:

Described expection resource load amount is stored in database, so that described service management system reads described expection resource load amount from described database, and generating resource deployment instruction according to described expection resource load amount, described resource management system is carried out resource deployment according to described resource deployment instruction.

4. according to arbitrary described resource load amount Forecasting Methodology in claim 1～3, it is characterized in that, described according to described business factor historical data sequence, generate expection business factor, comprising:

According to described business factor historical data sequence, draw the predicated error of at least two kinds of forecast models;

According to described business factor historical data sequence, adopt the forecast model of described predicated error minimum to calculate expection business factor.

5. according to arbitrary described resource load amount Forecasting Methodology in claim 1～3, it is characterized in that, described according to described business factor historical data sequence and resource load amount historical data sequence, set up the correlation function of business factor and resource load amount, comprising:

According to described business factor historical data sequence and resource load amount historical data sequence, adopt at least two function models to carry out the Function Fitting processing of described business factor and resource load amount, generate at least two candidate association functions;

According to described business factor historical data sequence and resource load amount historical data sequence, calculate the matching residual error that adopts each candidate association function according to default least-squares calculation algorithm;

The matching residual error of more each candidate association function, using the candidate association function of matching residual error minimum as described business factor and the correlation function of resource load amount.

6. a resource load amount Forecasting Methodology, is characterized in that, comprising:

Monitoring business factor;

The described business factor monitoring and monitoring time are stored in database, so that analysing and predicting system obtains business factor historical data sequence from described database, according to described business factor historical data sequence, generate expection business factor, according to described business factor historical data sequence and resource load amount historical data sequence, set up the correlation function of business factor and resource load amount, according to described correlation function, generate the expection resource load amount that described expection business factor is corresponding;

Wherein, the monitoring time that in described business factor historical data sequence, each business factor information is carried is identical with the monitoring time that each resource load amount information in described resource load amount historical data sequence is carried respectively, described resource load amount historical data sequence is obtained from described database by described analysing and predicting system, and described resource load amount information is that the resource load amount monitoring and monitoring time are together stored to the information in described database by resource management system.

7. an analysing and predicting system, is characterized in that, comprising:

The first acquisition module, for obtaining respectively business factor historical data sequence and resource load amount historical data sequence, wherein, the monitoring time that in described business factor historical data sequence, each business factor information is carried is identical with the monitoring time that each resource load amount information in described resource load amount historical data sequence is carried respectively;

The first processing module, for according to described business factor historical data sequence, generates expection business factor;

Set up module, for according to described business factor historical data sequence and resource load amount historical data sequence, set up the correlation function of business factor and resource load amount;

The second processing module, for according to described correlation function, generates the expection resource load amount that described expection business factor is corresponding.

8. analysing and predicting system according to claim 7, is characterized in that, also comprises:

The second acquisition module, for obtaining business factor information;

Judge module, for judging that monitoring time that described business factor information carries whether within the default unusual day period, if so, generates the first judgement information, otherwise, generate the second judgement information;

Execution module, for according to described the first judgement information, deposits described business factor information in business factor unusual day historical data sequence; According to described the second judgement information, deposit described business factor information in business factor normal day historical data sequence;

Described the second acquisition module, also for Gains resources charge capacity information;

Described judge module, also for judging that monitoring time that described resource load amount information carries whether within the default unusual day period, if so, generates the 3rd judgement information, otherwise, generate the 4th judgement information;

Described execution module, also, for according to described the 3rd judgement information, deposits described resource load amount information in resource load amount unusual day historical data sequence; According to described the second judgement information, deposit described resource load amount information in resource load amount normal day historical data sequence;

The described business factor historical data sequence that described the first acquiring unit obtains is unusual day historical data sequence of business factor, described resource load amount historical data sequence is unusual day historical data sequence of resource load amount, or, the described business factor historical data sequence that described the first acquiring unit obtains is the normal day historical data sequence of business factor, and described resource load amount historical data sequence is the normal day historical data sequence of resource load amount.

9. according to the analysing and predicting system described in claim 7 or 8, it is characterized in that, also comprise:

Memory module, for described expection resource load amount is stored to database, so that described service management system reads described expection resource load amount from described database, and generating resource deployment instruction according to described expection resource load amount, described resource management system is carried out resource deployment according to described resource deployment instruction.

10. a service operation supervisory system, is characterized in that, comprising:

Monitoring modular, for monitoring business factor;

Memory module, for the described business factor and the monitoring time that monitor are stored to database, so that analysing and predicting system obtains business factor historical data sequence from described database, according to described business factor historical data sequence, generate expection business factor, according to described business factor historical data sequence and resource load amount historical data sequence, set up the correlation function of business factor and resource load amount, according to described correlation function, generate the expection resource load amount that described expection business factor is corresponding;

Wherein, the monitoring time that in described business factor historical data sequence, each business factor information is carried is identical with the monitoring time that each resource load amount information in described resource load amount historical data sequence is carried respectively, described resource load amount historical data sequence is obtained from described database by described analysing and predicting system, and described resource load amount information is that the resource load amount monitoring and monitoring time are together stored to the information in described database by resource management system.

Images