CN108241864A - Server performance Forecasting Methodology based on multivariable grouping - Google Patents
Server performance Forecasting Methodology based on multivariable grouping Download PDFInfo
- Publication number
- CN108241864A CN108241864A CN201611213761.0A CN201611213761A CN108241864A CN 108241864 A CN108241864 A CN 108241864A CN 201611213761 A CN201611213761 A CN 201611213761A CN 108241864 A CN108241864 A CN 108241864A
- Authority
- CN
- China
- Prior art keywords
- computer system
- mts
- performance
- network server
- variable
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 238000000034 method Methods 0.000 title claims abstract description 21
- 238000004422 calculation algorithm Methods 0.000 claims abstract description 30
- 239000013598 vector Substances 0.000 claims description 23
- 239000011159 matrix material Substances 0.000 claims description 15
- HUTDUHSNJYTCAR-UHFFFAOYSA-N ancymidol Chemical compound C1=CC(OC)=CC=C1C(O)(C=1C=NC=NC=1)C1CC1 HUTDUHSNJYTCAR-UHFFFAOYSA-N 0.000 claims description 8
- 238000012545 processing Methods 0.000 claims description 4
- 238000004364 calculation method Methods 0.000 claims description 3
- 230000005540 biological transmission Effects 0.000 claims description 2
- 241001269238 Data Species 0.000 claims 8
- 238000005070 sampling Methods 0.000 claims 2
- 238000010845 search algorithm Methods 0.000 claims 2
- 230000015572 biosynthetic process Effects 0.000 abstract 1
- 238000003786 synthesis reaction Methods 0.000 abstract 1
- 230000015654 memory Effects 0.000 description 8
- 238000012549 training Methods 0.000 description 5
- 238000000714 time series forecasting Methods 0.000 description 4
- 238000005516 engineering process Methods 0.000 description 3
- 238000003646 Spearman's rank correlation coefficient Methods 0.000 description 2
- 230000008901 benefit Effects 0.000 description 2
- 230000008859 change Effects 0.000 description 2
- 238000004590 computer program Methods 0.000 description 2
- 238000010586 diagram Methods 0.000 description 2
- 230000006870 function Effects 0.000 description 2
- 238000011160 research Methods 0.000 description 2
- 238000012706 support-vector machine Methods 0.000 description 2
- 238000013528 artificial neural network Methods 0.000 description 1
- 239000003795 chemical substances by application Substances 0.000 description 1
- 238000004891 communication Methods 0.000 description 1
- 238000013500 data storage Methods 0.000 description 1
- 230000007423 decrease Effects 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 238000011478 gradient descent method Methods 0.000 description 1
- 238000009499 grossing Methods 0.000 description 1
- 238000012417 linear regression Methods 0.000 description 1
- 238000007726 management method Methods 0.000 description 1
- 238000005259 measurement Methods 0.000 description 1
- 230000007246 mechanism Effects 0.000 description 1
- 239000000203 mixture Substances 0.000 description 1
- YHXISWVBGDMDLQ-UHFFFAOYSA-N moclobemide Chemical compound C1=CC(Cl)=CC=C1C(=O)NCCN1CCOCC1 YHXISWVBGDMDLQ-UHFFFAOYSA-N 0.000 description 1
- 230000006855 networking Effects 0.000 description 1
- 239000007787 solid Substances 0.000 description 1
- 230000001360 synchronised effect Effects 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/23—Clustering techniques
-
- G—PHYSICS
- G16—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
- G16Z—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS, NOT OTHERWISE PROVIDED FOR
- G16Z99/00—Subject matter not provided for in other main groups of this subclass
Landscapes
- Engineering & Computer Science (AREA)
- Data Mining & Analysis (AREA)
- Theoretical Computer Science (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Bioinformatics & Computational Biology (AREA)
- Artificial Intelligence (AREA)
- Evolutionary Biology (AREA)
- Evolutionary Computation (AREA)
- Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Life Sciences & Earth Sciences (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
This disclosure relates to the server performance Forecasting Methodology based on multivariable grouping.The present invention proposes a kind of method for multi-step prediction enterprise servers performance.In highly dynamic Resources Sharing Environment, the Accurate Prediction of system performance index is the key that resource high-efficiency distribution.For the prediction of the performance indicators such as cpu load, researcher has proposed a series of prediction models, but is largely focused on single argument and short-term forecast field.This method has collected the History Performance Data of server, and converts it into Multivariate Time Series.When being predicted, the Multivariate Time Series of history are scanned for by k nearest neighbor algorithm, find the historical series closest with the state of current server performance and resource consumption.By the predicted value of the follow-up time sequence information synthesis future server performance condition of K most similar historical time sequences.
Description
Technical field
The present invention relates to the multivariable Server Load Prediction methods in information technology field.
Background technology
At present, there are many researchs to predict server performance by time series forecasting algorithm.
It is pre- that these research some have used classical linear regression, exponential smoothing or ARIMA models to carry out time series
It surveys, also some has used increasingly complex support vector machines (SVM), neural network and fuzzy logic algorithm.KNN algorithms also exist
It is used in time series forecasting.But the prediction of existing server performance both for single index, and have ignored
Correlation between the demand and index of multiple indexs prediction in practice；On the other hand, these predictions are directed in next step
Data, the data of following multiple step-lengths can not accurately be predicted.
Invention content
The technology of the present invention solves the problems, such as：
1. multistep performance prediction is obtained based on multidimensional k nearest neighbor method
Multistep performance information can be predicted by multidimensional k nearest neighbor method, expand the range of prediction.
2. it is grouped by dimension and improves predetermined speed
With the increase of time series dimension, carrying out the speed of time series forecasting can drastically decline.If prediction algorithm
Run time is long, then loses the ability predicted in real time.Predetermined speed is improved by variable grouping.
3. being grouped by dimension, the correlation between variable is remained
When there are many dimension of multivariate time series, directly carry out KNN predictions and produced instead due to the interference between variable
Raw very big error.It is grouped by dimension, remains the correlation between variable, and improve speed.
The technical solution of the present invention：In order to comprehensively utilize each indication information to improve the precision of prediction, the present invention
It is proposed a kind of server performance Forecasting Methodology for the K nearest neighbor algorithms being grouped based on multivariable.By analyzing the phase between each variable
Closing property is grouped variable, searches for similar historical sequence using K nearest neighbor methods on this basis, carries out multi-step prediction.
The advantages of the present invention over the prior art are that：
Multidimensional K arest neighbors method has been used to realize the multistep performance prediction to server.Meanwhile by variable by correlation,
Dependence is grouped so that the variable close relation in each group, the degree of correlation between the variables of difference group are relatively low or opposite
It is independent.Variable is grouped while interference raising precision of prediction is reduced, multivariate time series can also be greatly reduced
Dimension, so as to accelerate the search speed of KNN.
Description of the drawings
Fig. 1 KNN algorithm schematic diagrames.
Fig. 2 prediction algorithm flow charts.
Fig. 3 is the block diagram of the enterprise computer system of each embodiment according to the invention.
Specific embodiment
The present invention is divided into following steps：
S1 determines correlation matrix
The Multivariate Time Series that given length is T, dimension is d, we using related coefficient come weigh two variables it
Between correlation degree.This patent selection Spearman rank correlation coefficient (Spearman ' s rank correlation
Coefficient), it is the order statistical parameter that strong and weak nonparametric property is contacted between two variables of measurement.
To the sample that capacity is n, initial data Xi, YiIt is converted into level data xi, yi, correlation coefficient ρXYFor：
Wherein xi,yiRepresent initial data,Represent data mean value.
Due to the correlation between two variables be not necessarily it is synchronous, it is understood that there may be certain time delay, it is therefore desirable to examine
Consider influence of the time delay to related coefficient.For each pair of variable, algorithm can by when elongatedness its phase relation is calculated from 0 to MaxLag
Number, and therefrom choose the maximum value related coefficient final as the two, i.e.,：
Related coefficient of all d time serieses two-by-two between variable is calculated, the correlation matrix of a d × d is formed, comes
Portray connecting each other between this d variable.
S2 multivariables are grouped
Based on the correlation matrix of above-mentioned algorithm structure, this d variable is gathered into different by we using clustering algorithm
Group.
This patent has used neighbour's propagation clustering algorithm (Affinity Propagation, hereinafter referred to as AP).AP algorithms
It is a kind of clustering algorithm based on " message transmission " between data point.Unlike k-means algorithms and k-medoids algorithms,
AP algorithms do not need to the parameter of number either other description cluster numbers of specified cluster in advance, but by all data points all
As potential cluster centre.Since we are also unknown for the group result of these time series variables, AP algorithms pair
This patent solves the problems, such as more adaptability.
There are two types of types for the message transmitted in AP algorithms：Attraction Degree (responsibility) and degree of membership
(availability).The former is represented with r (i, k), for describing appropriateness of the point k as the cluster centre of data point i；Afterwards
Person is represented with a (i, k), for describing appropriateness of the point i selected elements k as its cluster centre.The two formula is as follows：
Wherein s (i, k) represents the similarity degree of i and k, can be obtained from similarity matrix.
AP algorithms receive similarity matrix and carry out starting algorithm as input, the Attraction Degree each put by iteration continuous renewal
With ownership angle value until convergence, generates several cluster centres, then remaining data point is assigned in suitable cluster.
The pseudocode of AP algorithms is as follows：
In packet by packet basis, future is predicted by finding history similar sequences.
S3 finds history similar sequences
In order to be better described, the method based on time series forecasting future multistep in the case of single argument is discussed first.
Give a limited equally spaced single time series xt, wherein t=(1,2 ..., T), it would be desirable in moment T
Multi-step prediction is carried out backward, and predicted value is expressed asWherein h=(1,2 ..., H), H are prediction step number.
First, in order to portray the feature of time series state change in the near future, we define one group of autoregressive feature
Pattern vector.These vectors are made of a series of continuous observations and equal length.In moment t, feature mode vector can
It is expressed asWherein m is feature mode length, also referred to as Embedded dimensions, is solid
Definite value.It should be noted that we can create a corresponding feature mode vector for each historical juncture, i.e., (t, t-1,
T-2 ..., t-m+1), therefore there is m-1 data item to be overlapped mutually in adjacent vector.The set of these vector compositions is referred to as m
Wei Lishixulieji (m-histories), m here tie up the time sequence spacing for referring to time series.
In next step, we calculate respectively all m dimension historical series set with finally it is observed that vectorThe distance between, we are measured using Euclidean distance here：
Finally, we sort each distance value being calculated, and it is a most similar with target feature vector therefrom to find out k
History vectors, and their follow-up data item is extracted respectively, combination generates predicted valueLast predicted value can be with table
It is shown as the weighted sum of k neighbours.Common combination has the weight (distance- of simple average or distribution based on distance
Based weights) summation.In order to more accurately describe the relationship of k neighbours and predicted value, we use gradient descent algorithm,
The weight of each neighbour is obtained by training：
Wherein, neighj,hFor the successor value of j-th of neighbour, wjThe respective weights obtained for training.
In the case of multivariable, the feature mode vector sum m-histories in above-mentioned single argument k-NN algorithms is expanded
Exhibition continues to use variable representation therein to hyperspace.
For the multivariate time series (MTS) of d dimensions, at the T moment, in order to predict the future value of the MTS, we define target
EigenmatrixWhereinRepresent in moment T, the target signature of dimension l to
Amount.Similarly, distance function be defined as each dimension target feature vector corresponded to m-histories dimension vector
Sum of the distance：
By the search to m-histories, we find out and the eigenmatrix under MTS current statesMost like k
A neighbour, each neighbour are m × d rank matrixes, are made of the feature mode vector that d length is m.For each dimension
For, this subsequent data point of k neighbour in historical series collectively constitutes R-matrix neigh：
So as to which MTS is by the data of respective dimensions in R-matrix to each dimension predicted value of moment T+h in moment T
Point weighted array forms：
S4 right value updates
There is certain potential trend in view of multivariate time series.Over time, predicted value and neighbour's value it
Between relationship can also gradually change, weights also can be more and more improper, therefore just needs to adjust weight in time with data variation
If Traint=[xt-1,xt-2,...,xt-n] represent that the training sample set that the size of t moment is n (does not include xt), xt
Represent the actual value of t moment, ytRepresent the predicted value of t moment.We are according to trained weights before, by TraintIt predicts
yt, then in prediction yt+1Before, historical series can be updated, add in xt, training sample set also can be with elapsing, similar to a movement backward
Window, so as to Traint+1=[xt,xt-1,...,xt-n+1], it is upper it is primary on the basis of train weights again, predict again later
yt+1, and so on.
In this way, weights can be on the basis of initially training collection, as the numerical value predicted later constantly updates adjustment；And it trains
Distant historical series are also gradually eliminated in the passage of sample set, and prediction model is made to adapt to the newest variation rule of time series
Rule.The characteristics of by gradient descent method, it is not very big to update the calculation amount needed for weights every time, is not had substantially to program runtime
Have an impact.
The details of the implementation of the embodiment of the present invention presented below.An exemplary enterprise calculation is shown in Fig. 3
Machine system 10, wherein can be for the use of the present invention.The enterprise computer system 10 shown in Fig. 3 includes and wide area network
(WAN) several LANs (LAN) 12 of 14 interconnection.Each LAN 12 may include several client computers 16 and several networks
Server 18.For example, depending on implementation, network server 18 can be in its LAN 12 or from other LAN
12 16 trustship of client computer (host) computer resource, such as computer program, data, storage device and printer.
According to various embodiments, multivariate time series of the resources computer system 20 based on network server 18
(MTS) data perform resources for enterprise, and wherein MTS data are stored in database computer system 22.For signal
Purpose, resources computer system 20 and MTS Database Systems 22 are shown connected to WAN 14 in figure 3, although they
In one or two can be included in one in shown LAN 12.They can also be connected in the network of the enterprise
Different LAN 12 and WAN 14.Resources computer system 20 can be implemented as the computer of one or several interconnection
Equipment, such as server, mainframe, work station and/or any other suitable computer equipment.Resources computer system
The such computer equipment of each of 20 can include one or more processors 24 and one or more memory cells 26.
Memory cell 26 can include base computer storage device (such as RAM and ROM) and second computer storage device (such as
HDD, SSD, flash memory).It as shown in Figure 3 and is described further below, processor 24 can include microprocessor, for holding
Row is stored in the computer instruction (such as software) in memory cell 26, such as variable grouping module 30 and K arest neighbors (KNN)
Search module 32.For illustration purposes, shown resources computer system 20 includes an only computer, and only one
A processor 24 and a memory cell 26 are shown, and the present invention is not so limited, and provide though it should be recognized that
Source prediction computer system 20 can zoom in and out as needed.
MTS databases 22 store the time series computer dosage and hardware statistics of the network server 18 of business system 10
Data.Statistical data can include the value for being related to several variables of each user resources request, such as：
Ask the user name of the user of resource
At the beginning of request
The end time of request
The total time of request
The activity time of request
Requested processing or resource
The ID of the network server of request is handled
The geometric position of the network server of request is handled
- CPU usage amounts
Background memory (such as RAM) usage amount of network server
The disk IO of network server (to total read and write operation of disk storage or other secondary computer memories)
MTS databases 22 can be embodied as having one or more database servers, the one or more as needed
Database server operation data base management system (DBMS) software and including suitable RAID disk array and/or any other conjunction
Suitable data storage mechanism.Csv file and/or any other suitable data format can be used by dosage and hardware statistics data
MTS databases 22 are sent to from network server 18.Each network server 18 can regularly statistical data be sent to by it
One or more of network server 18 in MTS databases 22 and/or LAN 12 can be collected more in LAN 12
Collected statistical data is simultaneously sent to MTS databases 22 by the statistical data of a network server 18.MTS databases 22 can quilt
It is embodied as such as Oracle or SQL database or any other suitable database.
When the software of performance variable grouping module 30, the processor of resources computer system 20 is as described above
Variable is grouped.When performing k-NN search modules 32, the processor of resources computer system 20 performs the above
K-NN search, including prediction and right value update.
As described above, resources computer system 20 can be based on the user job load forecast forecast for specific
Period, server 18 for enterprise right quantity.Resources computer system 20 can be by these server counts
Amount recommends to transmit the network server 40 that most network server 18 serves as agency (see Fig. 3).Based on from resources computer
The decision that system 20 is transmitted, the proxy server 40 can determine the how many servers 18 in LAN 12 at each moment
It should be opened (fully active, operating mode) and/or low-power mode (such as idle or sleep mode) should be placed into simultaneously
Correspondingly instruct these servers 18.In this way, when based on one in the dosage and/or load model network server 18 forecast
When not forecast a bit to be required, they can be placed into low-power mode.For example, with reference to figure 3, resources computer system 20
It can determine the quantity of the forecast for the required network server 18 of special time period.This forecast can be for one
Network server 18 in LAN 12 is either for the network server 18 across multiple LAN 12.Resources computer system
20 can be sent to this quantity proxy server 40, and proxy server 40 can instruct in its LAN 12 and/or in other LAN 12
Each server 18 appropriate at the time of in work, (user resources demand can be handled) in high-power mode or located
(user resources demand cannot be handled) in low-power mode.Network server 18 can correspondingly take commanded power mould
Formula.In this way, at the time of the network resource requirement of forecast is low, several network servers 18 can be placed into low-power mode to save
Energy and relevant cost.Conversely, at the time of the network resource requirement of forecast is high, sufficient amount of network server 18 will prepare
User expected from benefit reason asks load.
It should be noted that such as Fig. 3 illustrates only the component for the enterprise computer system 10 for being enough to understand the aspect of the present invention.
It should be appreciated that enterprise computer system 10 can include the use of wired (such as Ethernet on twisted-pair cable) or wireless
LAN, WAN, MAN of several network interconnections of (such as Wi-Fi) communication link etc..Network server 18 can handle institute of enterprise
It is required that, a large amount of and different types of resource, and client computer 16 can be the enabling network of any suitable type
End-user computer device, such as laptop, personal computer, tablet computer, smart phone etc..Resource is pre-
Surveying computer system 20 can be realized by the computer equipment of one or more networkings.It is wrapped in resources computer system 20
In the case of including multiple computer equipments, they can be by network interconnections such as one or more LAN, WAN, MAN.In addition, enterprise
Computer system 10 can include additional Agent Computer 40 with provisioning server 18.
Software module 30,32 can realize in one write with any suitable computer language (such as Python) or
In multiple computer programs, so as to which processor 24 when the software program of 24 execution module 30,32 of processor, be made to perform this paper institutes
The function of module 30,32 of stating.For example, include general purpose microprocessor for the suitable processor 24 of execute instruction program
Both with special microprocessor.In addition, any machine element as described herein can include single processor or multiple processing
Device.Processor 24 is received from read-only memory or random access memory or the instruction and data of both.
Claims (14)
1. a kind of computer implemented method that network resource requirement is forecast for enterprise computer system, wherein the enterprise calculation
Machine system includes multiple network servers of user's trustship computer resource for the enterprise computer system, the method packet
It includes：
By computer database system, multivariate time series (MTS) performance data of the multiple network server is received,
Described in MTS performance datas include a series of for prior sample moment, multiple d for the multiple network server
The data of energy variable；
It, will be in the MTS performance datas by the computer system programmed to communicate with the computer database system
Variable is grouped into two or more set of variables so that each in the performance variable in the MTS performance datas
Belong to a set of variables；
By the computer system programmed, by calculate at one or more future time range steps to the variable
Prediction, to calculate the prediction of the future workload of the network server of the enterprise computer system, wherein calculating
The prediction includes：
Using the k nearest neighbor search algorithms for being applied to the two or more set of variables, find relative to the MTS performance numbers
According to normal condition k arest neighbors；And calculate the weighted average of the k arest neighbors；And
By the computer system programmed, based on the prediction calculated, the recommended amount of network server is determined, this is described
Enterprise needs in the operating mode, with handle in one or more of future time range steps each it is following when
Between the enterprise computer system at range step the user resource request network server recommended amount.
2. the method as described in claim 1, wherein the performance variable of the multiple network server includes instruction at least
The variable of the following contents：
Cpu load；
Base computer memory usage amount；And
Input/output (IO) operation of section secondary computer storage device per unit time.
3. the method as described in claim 1, wherein the step of variable is grouped includes：
By the computer system programmed, correlation matrix of the instruction per the correlation between a pair of of performance variable is calculated；
And
By the computer system programmed, the set of variables is determined based on the correlation matrix using clustering algorithm.
4. method as claimed in claim 3, wherein the correlation matrix includes Spearman correlation matrix.
5. method as claimed in claim 4, wherein the clustering algorithm includes neighbour's propagation clustering algorithm.
6. the method as described in claim 1, wherein finding relative to described in the normal condition of the MTS performance datas
The step of k arest neighbors, includes：
The MTS number of the vector sum representative for the normal condition for representing the MTS at previous sampling instant is calculated respectively
According to multiple vectors in the distance between each vector；And
Determining vector for the prior sample moment, relative to the normal condition for representing the MTS has minimum range
K vector.
7. method as claimed in claim 6, include calculating Euclidean distance wherein calculating distance.
8. the method as described in claim 1 further includes：
By the computer system transmission data programmed, the data indicate that the enterprise is needed in operating mode
To handle at least one of future time range step future time range step, described enterprise computer system
The recommended amount of the network server of the resource request of the user of system；And
The network is instructed to take by the one or more proxy computer systems to communicate with the computer system programmed
Business device so that the network server of the recommended amount is in operating mode to handle at least one future time
The resource request of range step, described user.
9. a kind of system for being used to forecast network resource requirement for enterprise computer system, wherein the enterprise computer system packet
Multiple network servers of user's trustship computer resource for the enterprise computer system are included, the system comprises：
The computer database system to communicate with the multiple network server, for store multiple network servers it is polynary when
Between sequence (MTS) performance data, wherein the MTS performance datas include for a series of prior sample moment, for multiple
The data of multiple d performance variables of network server；
The computer system programmed to communicate with the computer database system, wherein the computer system programmed
The step of being programmed to by performing comprising following operation predicts the network server of the enterprise computer system
Future workload：
Variable in the MTS performance datas is grouped into two or more set of variables so that in the MTS performance datas
The performance variable in each belong to a set of variables；
The prediction at one or more future time range steps to the variable is calculated, the prediction indicates the network clothes
The workload of business device, wherein being calculated the step of prediction is by including operations described below：
Using the k nearest neighbor search algorithms for being applied to the two or more set of variables, find relative to the MTS performance numbers
According to normal condition k arest neighbors；And
Calculate the weighted average of the k arest neighbors；And
Based on the prediction calculated, determine the enterprise need in operating mode, with processing it is one or more of not
Carry out the money of the user of the enterprise computer system at each future time range step in time range step-length
The quantity of the network server of source request；And
At least one proxy computer system to communicate with the computer system programmed and the multiple network server,
Wherein described at least one proxy computer system is used for：
Receive data from the computer system that has programmed, the data indicate the enterprise need in operating mode with
Processing is for the enterprise computer system of at least one of future time range step future time range step
The user the resource request network server recommended amount；
Instruct the network server so that the network server of the recommended amount is in operating mode to handle for extremely
The resource request of the user of a few future time range step.
10. system as claimed in claim 9, wherein the performance variable of the multiple network server includes at least indicating
The variable of the following contents：
Cpu load；
Base computer memory usage amount；And
Input/output (IO) operation of section secondary computer storage device per unit time.
11. system as claimed in claim 9, wherein the step of variable is grouped includes：
By the computer system programmed, correlation matrix of the instruction per the correlation between a pair of of performance variable is calculated；
And
By the computer system programmed, the set of variables is determined based on the correlation matrix using clustering algorithm.
12. system as claimed in claim 11, wherein the correlation matrix includes Spearman correlation matrix.
13. system as claimed in claim 11, wherein the clustering algorithm includes neighbour's propagation clustering algorithm.
14. system as claimed in claim 9, wherein finding the institute of the normal condition relative to the MTS performance datas
The step of stating k arest neighbors includes：
The MTS number of the vector sum representative for the normal condition for representing the MTS at previous sampling instant is calculated respectively
According to multiple vectors in the distance between each vector；And
Determining vector for the prior sample moment, relative to the normal condition for representing the MTS has minimum range
K vector.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201611213761.0A CN108241864A (en) | 2016-12-26 | 2016-12-26 | Server performance Forecasting Methodology based on multivariable grouping |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201611213761.0A CN108241864A (en) | 2016-12-26 | 2016-12-26 | Server performance Forecasting Methodology based on multivariable grouping |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN108241864A true CN108241864A (en) | 2018-07-03 |
Family
ID=62704559
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201611213761.0A Pending CN108241864A (en) | 2016-12-26 | 2016-12-26 | Server performance Forecasting Methodology based on multivariable grouping |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN108241864A (en) |
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111625440A (en) * | 2020-06-04 | 2020-09-04 | 中国银行股份有限公司 | Method and device for predicting performance parameters |
| CN111773663A (en) * | 2020-07-09 | 2020-10-16 | 网易（杭州）网络有限公司 | Game server merging effect prediction method, device, equipment and storage medium |
| CN112116480A (en) * | 2019-06-20 | 2020-12-22 | 财付通支付科技有限公司 | Virtual resource determination method and device, computer equipment and storage medium |
| CN112783740A (en) * | 2020-12-30 | 2021-05-11 | 科大国创云网科技有限公司 | Server performance prediction method and system based on time series characteristics |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN1605060A (en) * | 2001-12-17 | 2005-04-06 | 国际商业机器公司 | Automatic data interpretation and implementation using performance capacity management framework over many servers |
| CN101410806A (en) * | 2006-03-27 | 2009-04-15 | 国际商业机器公司 | Approximating resource consumption of a computer system |
| CN103778474A (en) * | 2012-10-18 | 2014-05-07 | 华为技术有限公司 | Resource load capacity prediction method, analysis prediction system and service operation monitoring system |
| CN104010029A (en) * | 2014-05-12 | 2014-08-27 | 上海交通大学 | Distributed computing environment performance predicting method based on transverse and longitudinal information integration |
| US20150026108A1 (en) * | 2013-03-15 | 2015-01-22 | Citrix Systems, Inc. | Managing Computer Server Capacity |
-
2016
- 2016-12-26 CN CN201611213761.0A patent/CN108241864A/en active Pending
Patent Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN1605060A (en) * | 2001-12-17 | 2005-04-06 | 国际商业机器公司 | Automatic data interpretation and implementation using performance capacity management framework over many servers |
| CN101410806A (en) * | 2006-03-27 | 2009-04-15 | 国际商业机器公司 | Approximating resource consumption of a computer system |
| CN103778474A (en) * | 2012-10-18 | 2014-05-07 | 华为技术有限公司 | Resource load capacity prediction method, analysis prediction system and service operation monitoring system |
| US20150026108A1 (en) * | 2013-03-15 | 2015-01-22 | Citrix Systems, Inc. | Managing Computer Server Capacity |
| CN105051693A (en) * | 2013-03-15 | 2015-11-11 | 思杰系统有限公司 | Managing computer server capacity |
| CN104010029A (en) * | 2014-05-12 | 2014-08-27 | 上海交通大学 | Distributed computing environment performance predicting method based on transverse and longitudinal information integration |
Non-Patent Citations (1)
| Title |
|---|
| LANG WANG ETC.: "A Prediction based Capacity Planning Strategy for Virtual Servers", 《2015 IEEE INTERNATIONAL CONFERENCE ON DATA SCIENCE AND DATA INTENSIVE SYSTEMS》 * |
Cited By (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112116480A (en) * | 2019-06-20 | 2020-12-22 | 财付通支付科技有限公司 | Virtual resource determination method and device, computer equipment and storage medium |
| CN111625440A (en) * | 2020-06-04 | 2020-09-04 | 中国银行股份有限公司 | Method and device for predicting performance parameters |
| CN111773663A (en) * | 2020-07-09 | 2020-10-16 | 网易（杭州）网络有限公司 | Game server merging effect prediction method, device, equipment and storage medium |
| CN111773663B (en) * | 2020-07-09 | 2024-02-23 | 网易（杭州）网络有限公司 | Game server merging effect prediction method, device, equipment and storage medium |
| CN112783740A (en) * | 2020-12-30 | 2021-05-11 | 科大国创云网科技有限公司 | Server performance prediction method and system based on time series characteristics |
| CN112783740B (en) * | 2020-12-30 | 2022-11-18 | 科大国创云网科技有限公司 | Server performance prediction method and system based on time series characteristics |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| US11481653B2 (en) | Predictive asset optimization for computer resources | |
| US11599393B2 (en) | Guaranteed quality of service in cloud computing environments | |
| CN107404523A (en) | Cloud platform adaptive resource dispatches system and method | |
| Rahman et al. | Replica selection strategies in data grid | |
| Rodrigues et al. | Helping HPC users specify job memory requirements via machine learning | |
| CN108241864A (en) | Server performance Forecasting Methodology based on multivariable grouping | |
| CN118069380A (en) | Computing power resource processing method | |
| Li et al. | Efficient response time predictions by exploiting application and resource state similarities | |
| Qian et al. | An enhanced genetic algorithm for constrained knapsack problems in dynamic environments | |
| Li et al. | Mining performance data for metascheduling decision support in the grid | |
| Chen et al. | Silhouette: Efficient cloud configuration exploration for large-scale analytics | |
| JP2021005387A (en) | Predictive asset optimization for computer resources | |
| JP2021005387A5 (en) | ||
| CN108241533B (en) | Resource pool future load generation method based on prediction and hierarchical sampling | |
| Sun et al. | Aledar: An attentions-based encoder-decoder and autoregressive model for workload forecasting of cloud data center | |
| CN113191540A (en) | Construction method and device of industrial link manufacturing resources | |
| Rossi et al. | Clustering-Based Numerosity Reduction for Cloud Workload Forecasting | |
| Shao et al. | ℓp‐Norm Multikernel Learning Approach for Stock Market Price Forecasting | |
| Gao et al. | Energy-efficient buffer and service rate allocation in manufacturing systems using hybrid machine learning and evolutionary algorithms | |
| Dou et al. | Intelligent business resources deployment over industrial internet | |
| Zeng | A clustering method of combining grid and genetic algorithm in wireless sensor networks | |
| Huang et al. | The Implementation of Multiobjective Flexible Workshop Scheduling Based on Genetic Simulated Annealing‐Inspired Clustering Algorithm | |
| Elomari et al. | DFS Response Time Prediction Using the Techniques of “Deep Learning” | |
| Xia et al. | Deep&Cross Network for Software-Intensive System Fault Prediction | |
| Wu et al. | Storage and Query Indexing Methods on Big Data |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| REG | Reference to a national code |
Ref country code: HK Ref legal event code: DE Ref document number: 1257207 Country of ref document: HK |
|
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20180703 |
|
| RJ01 | Rejection of invention patent application after publication |