CN110109800A - A kind of management method and device of server cluster system - Google Patents
A kind of management method and device of server cluster system Download PDFInfo
- Publication number
- CN110109800A CN110109800A CN201910286423.7A CN201910286423A CN110109800A CN 110109800 A CN110109800 A CN 110109800A CN 201910286423 A CN201910286423 A CN 201910286423A CN 110109800 A CN110109800 A CN 110109800A
- Authority
- CN
- China
- Prior art keywords
- transaction data
- moment
- model
- data amount
- server
- 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
- 238000007726 management method Methods 0.000 title claims abstract description 38
- 238000012545 processing Methods 0.000 claims abstract description 72
- 238000000034 method Methods 0.000 claims abstract description 32
- 238000012549 training Methods 0.000 claims description 42
- 238000009499 grossing Methods 0.000 claims description 22
- 230000000737 periodic effect Effects 0.000 claims description 12
- 238000004590 computer program Methods 0.000 claims description 10
- 238000003860 storage Methods 0.000 claims description 5
- 238000004364 calculation method Methods 0.000 claims description 2
- 238000009826 distribution Methods 0.000 description 21
- 230000007246 mechanism Effects 0.000 description 20
- 238000010586 diagram Methods 0.000 description 13
- 230000008569 process Effects 0.000 description 7
- 230000004048 modification Effects 0.000 description 5
- 238000012986 modification Methods 0.000 description 5
- 230000006854 communication Effects 0.000 description 4
- 230000000694 effects Effects 0.000 description 4
- 238000004891 communication Methods 0.000 description 3
- 230000006870 function Effects 0.000 description 3
- 239000013307 optical fiber Substances 0.000 description 3
- 125000002015 acyclic group Chemical group 0.000 description 2
- 235000013399 edible fruits Nutrition 0.000 description 2
- 238000005516 engineering process Methods 0.000 description 2
- 230000003466 anti-cipated effect Effects 0.000 description 1
- 230000005540 biological transmission Effects 0.000 description 1
- 230000008859 change Effects 0.000 description 1
- 230000003111 delayed effect Effects 0.000 description 1
- 238000010801 machine learning Methods 0.000 description 1
- 238000004519 manufacturing process Methods 0.000 description 1
- 230000003287 optical effect Effects 0.000 description 1
- 230000009467 reduction Effects 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F11/00—Error detection; Error correction; Monitoring
- G06F11/30—Monitoring
- G06F11/3003—Monitoring arrangements specially adapted to the computing system or computing system component being monitored
- G06F11/3006—Monitoring arrangements specially adapted to the computing system or computing system component being monitored where the computing system is distributed, e.g. networked systems, clusters, multiprocessor systems
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F11/00—Error detection; Error correction; Monitoring
- G06F11/30—Monitoring
- G06F11/3051—Monitoring arrangements for monitoring the configuration of the computing system or of the computing system component, e.g. monitoring the presence of processing resources, peripherals, I/O links, software programs
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q40/00—Finance; Insurance; Tax strategies; Processing of corporate or income taxes
- G06Q40/04—Trading; Exchange, e.g. stocks, commodities, derivatives or currency exchange
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Computing Systems (AREA)
- Business, Economics & Management (AREA)
- General Physics & Mathematics (AREA)
- Accounting & Taxation (AREA)
- General Engineering & Computer Science (AREA)
- Quality & Reliability (AREA)
- Finance (AREA)
- Development Economics (AREA)
- Economics (AREA)
- Marketing (AREA)
- Strategic Management (AREA)
- Technology Law (AREA)
- General Business, Economics & Management (AREA)
- Mathematical Physics (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The embodiment of the invention discloses a kind of management method of server cluster system and devices, wherein method includes: to obtain N number of moment of multiple servers before the first moment corresponding transaction data amount in server cluster system, and it is predicted to obtain the corresponding transaction data amount of object time according to corresponding transaction data amount of N number of moment, and then adjust the configuration for handling the quantity of the server of transaction data in server cluster system and/or handling the server of transaction data before object time.In the embodiment of the present invention, by using the transaction data amount of the transaction data amount prediction object time before object time, it can make server cluster system that the management to server cluster system can be completed before object time arrival, so as to improve the flexibility of management, the efficiency of server cluster system processing transaction data is improved.
Description
Technical field
The present invention relates to field of computer technology more particularly to the management methods and device of a kind of server cluster system.
Background technique
Multiple servers usually can be set in server cluster system, it, can be with if receiving trading processing request
Using any one in multiple servers or any multiple servers handle transaction data.By in server cluster
Multiple servers are set in system, server cluster system can be made to handle more transaction data, improve server cluster
System handles the ability of transaction data, i.e. server cluster system can be applied in the scene of big transaction data amount.
In practical applications, however, it is determined that the ability of server cluster system processing transaction data and current time are to be processed
Transaction data amount mismatches, then may need to be managed server cluster system, for example carries out to server cluster system
Dilatation or capacity reducing.Existing a kind of management method are as follows: after receiving trading processing request, obtain trading processing and request corresponding friendship
Easy data volume, however, it is determined that server cluster system is less than trading processing request in the ability of current time processing transaction data and corresponds to
Transaction data amount, then can to server cluster system carry out capacity reducing, such as can reduce processing transaction data server
Quantity or reduce processing transaction data server hardware configuration;If it is determined that server cluster system is at current time
The ability for managing transaction data, which is greater than trading processing, requests corresponding transaction data amount, then can expand server cluster system
Hold, for example the quantity of the server of processing transaction data can be increased or improve the hardware of server of processing transaction data and matched
It sets.However, aforesaid way can be managed to server cluster system after getting trading processing request, it is possible to meeting
Since current network problem causes management process to be delayed, i.e., when trading processing request arrives, server cluster system may also not
Management is completed, consequently, it is possible to the efficiency for handling server cluster system to transaction data is poor.
To sum up, the management method for needing a kind of server cluster system at present, to improve the place of server cluster system
Manage efficiency.
Summary of the invention
The embodiment of the present invention provides the management method and device of a kind of server cluster system, to improve server cluster
The treatment effeciency of system.
In a first aspect, a kind of management method of server cluster system provided in an embodiment of the present invention, comprising:
Obtain N number of moment corresponding number of deals of multiple servers before the first moment in server cluster system
It predicts to obtain the corresponding transaction data amount of object time according to amount, and according to N number of moment corresponding transaction data amount, wherein
At the time of the object time is after first moment or first moment;Further, according to the object time
Corresponding transaction data amount adjusts the service that transaction data is handled in the server cluster system before the object time
The configuration of the server of the quantity and/or processing transaction data of device.
In above-mentioned technical proposal, by using the transaction data of the transaction data amount prediction object time before object time
Amount can make server cluster system that the management to server cluster system can be completed before object time arrival, from
And the flexibility of management can be improved;And by way of being in advance managed server cluster system, it can to manage
Server cluster system afterwards can be handled in time the corresponding transaction data of object time in object time, so as to improve clothes
The efficiency of device group system of being engaged in processing transaction data.
Optionally, described to predict to obtain the corresponding transaction of object time according to N number of moment corresponding transaction data amount
Data volume, comprising: obtain the first initial model, and corresponding transaction data amount of N number of moment is used to train at the beginning of described first
Beginning model obtains the first model, wherein first initial model is multinomial model;Further, using first mould
Type is predicted to obtain the corresponding first transaction data amount of the object time, and determines institute according at least to the first transaction data amount
State the corresponding transaction data amount of object time.
In above-mentioned technical proposal, by being predicted based on multinomial model the corresponding transaction data amount of object time,
The transaction data amount that prediction can be made to obtain is more accurate;And it is more by using the corresponding transaction data amount adjustment of N number of moment
Parameter in item formula model, the parameter for the multinomial model that training can be made to obtain is more accurate, i.e., multinomial model is pre-
Survey effect is preferable, so as to further improve the accuracy for the transaction data amount that prediction obtains.
Optionally, described to determine the corresponding transaction data of the object time according at least to the first transaction data amount
Amount, comprising: if it is determined that corresponding transaction data amount of N number of moment belongs to aperiodicity data, then the second initial model is obtained,
And using N number of moment corresponding transaction data amount training second initial model, the second model is obtained;Further,
The corresponding second transaction data amount of the object time is obtained using second model prediction, and according to first number of deals
According to amount and the second transaction data amount, the corresponding transaction data amount of the object time is determined, second initial model is
Second multinomial exponential smoothing model;Or, however, it is determined that N number of moment, corresponding transaction data amount belonged to periodic data, then obtained
Third initial model, and using N number of moment corresponding transaction data amount training third initial model, obtain third mould
Type;Further, the corresponding third transaction data amount of the object time is obtained using the third model prediction, and according to institute
The first transaction data amount and the third transaction data amount are stated, determines the corresponding transaction data amount of the object time, described
Three initial models are Three-exponential Smoothing model.
In above-mentioned technical proposal, by determining whether transaction data amount is periodic data, can choose different models
(i.e. the second model and third model) predicts periodic transaction data amount and acyclic transaction data amount respectively,
So that determining that the obtained corresponding transaction data amount of object time is more accurate, and then can be based on the transaction
The server cluster system of data volume management is more in line with actual conditions;And object time pair determined by above-mentioned technical proposal
The transaction data amount answered is obtained by two model predictions, Individual forecast knot of the prediction result compared to two models
It is more accurate for fruit, so as to more improve ground management server group system based on the transaction data amount.
Optionally, in N number of moment with first moment it is close at the time of be the second moment;It is described to use the N
A moment corresponding transaction data amount training second initial model, obtains the second model, comprising: use N number of moment
In N-1 moment in addition to second moment corresponding transaction data amount training second initial model, obtain described
Second moment corresponding second model;According to second moment corresponding second model and second moment corresponding transaction
Data volume training second moment corresponding second model, obtains second model;Alternatively, it is described using it is described N number of when
The corresponding transaction data amount training third initial model is carved, third model is obtained, comprising: using being removed in N number of moment
The N-1 moment corresponding transaction data amount training third initial model other than second moment, obtains described second
Moment corresponding third model;According to second moment corresponding third model and second moment corresponding transaction data
Amount training second moment corresponding third model, obtains the third model.
In above-mentioned technical proposal, the second model or third model are determined by using the mode of iteration, it can be with real-time update
Model parameter in second model or third model, so that the prediction effect of the second model or third model is more preferable, precision
It is higher, improve the accuracy of prediction.
Optionally, described according to the first transaction data amount and the second transaction data amount, when determining the target
Carve corresponding transaction data amount, comprising: obtain corresponding first weight of the first model and the corresponding power of second model
Weight, and according to the first transaction data amount, corresponding first weight of first model, the second transaction data amount and institute
The corresponding weight of the second model is stated, determines the corresponding transaction data amount of the object time;
It is described according to the first transaction data amount and the third transaction data amount, determine that the object time is corresponding
Transaction data amount, comprising: acquisition corresponding second weight of the first model and the corresponding weight of the third model, and according to
The first transaction data amount, corresponding second weight of first model, the third transaction data amount and the third mould
The corresponding weight of type determines the corresponding transaction data amount of the object time.
In above-mentioned technical proposal, the weight of different models can be configured according to the actual situation, so that base
It is more accurate in the transaction data amount that Weight prediction obtains, it is more in line with actual conditions;That is, by setting weight, it can
So that the application scenarios of above-mentioned technical proposal are more extensive.
Optionally, described according to the corresponding transaction data amount of the object time, institute is adjusted before the object time
The configuration for handling the quantity of the server of transaction data in server cluster system and/or handling the server of transaction data is stated,
It include: the quantity and processing number of deals for handling the server of transaction data when obtaining for the second moment in the server cluster system
According to server configuration, and the server of transaction data is handled according to second moment when server cluster system
Quantity and processing transaction data server configuration, determine that second moment corresponding server cluster system can be located
The transaction data amount of reason, wherein at the time of second moment is nearest before first moment;Further, if it is described
The transaction data amount that second moment corresponding server cluster system can be handled is greater than the corresponding number of deals of the object time
According to amount, then increase the quantity that the server of transaction data is handled in the server cluster system before the object time
And/or the configuration that at least one server of transaction data is handled in the server cluster system is improved, if when described second
It carves the transaction data amount that corresponding server cluster system can be handled and is less than the corresponding transaction data amount of the object time, then
It reduces the quantity for handling the server of transaction data in the server cluster system and/or reduces the server cluster system
The configuration of at least one server of middle processing transaction data.
In above-mentioned technical proposal, by that will predict the corresponding transaction data amount of obtained object time and current server collection
Group system processing capacity compare, can before object time by the processing capacity of server cluster system be adjusted to
The transaction data amount of prediction matches so that server cluster system can when object time arrives timely processing target
Moment corresponding real trade data volume improves the efficiency of server cluster system processing transaction data.
Second aspect, the embodiment of the present invention provide a kind of managing device of server cluster system, which includes:
Module is obtained, for obtaining N number of moment of multiple servers before the first moment point in server cluster system
Not corresponding transaction data amount;
Prediction module, for predicting to obtain the corresponding friendship of object time according to N number of moment corresponding transaction data amount
Easy data volume;At the time of the object time is after first moment or first moment;
Management module, for being adjusted before the object time according to the corresponding transaction data amount of the object time
The quantity of the server of transaction data is handled in the server cluster system and/or handles matching for the server of transaction data
It sets.
Optionally, the prediction module is used for: obtaining the first initial model, first initial model is polynomial module
Type;Using N number of moment corresponding transaction data amount training first initial model, the first model is obtained;Using described
First model prediction obtains the corresponding first transaction data amount of the object time, and according at least to the first transaction data amount
Determine the corresponding transaction data amount of the object time.
Optionally, the prediction module is used for: if it is determined that corresponding transaction data amount of N number of moment belongs to aperiodicity
Data then obtain the second initial model, using N number of moment corresponding transaction data amount training second initial model,
Obtain the second model;The corresponding second transaction data amount of the object time is obtained using second model prediction, and according to
The first transaction data amount and the second transaction data amount determine the corresponding transaction data amount of the object time, described
Second initial model is second multinomial exponential smoothing model;Or, however, it is determined that N number of moment, corresponding transaction data amount belonged to the period
Property data, then obtain third initial model, use N number of moment corresponding transaction data amount training third introductory die
Type obtains third model;The corresponding third transaction data amount of the object time is obtained using the third model prediction, according to
The first transaction data amount and the third transaction data amount determine the corresponding transaction data amount of the object time, described
Third initial model is Three-exponential Smoothing model.
Optionally, in N number of moment with first moment it is close at the time of be the second moment;The prediction module is used
In: use the N-1 moment corresponding transaction data amount training described second in N number of moment in addition to second moment
Initial model obtains second moment corresponding second model;According to second moment corresponding second model and described
Corresponding second model of corresponding transaction data amount training second moment at second moment, obtains second model;Alternatively,
At the beginning of using the N-1 moment corresponding transaction data amount training third in N number of moment in addition to second moment
Beginning model obtains second moment corresponding third model;According to second moment corresponding third model and described
Corresponding third model of corresponding transaction data amount training second moment at two moment, obtains the third model.
Optionally, the prediction module is used for: obtaining corresponding first weight of the first model and second model
Corresponding weight;According to the first transaction data amount, corresponding first weight of first model, second transaction data
Weight corresponding with second model is measured, determines the corresponding transaction data amount of the object time;Alternatively, obtaining described first
Corresponding second weight of model and the corresponding weight of the third model;According to the first transaction data amount, first mould
Corresponding second weight of type, the third transaction data amount and the corresponding weight of the third model, determine the object time
Corresponding transaction data amount.
Optionally, the management module is used for: handling number of deals in the server cluster system when obtaining for the second moment
According to server quantity and processing transaction data server configuration;Second moment be first moment before most
At the time of close;The quantity and processing for handling the server of transaction data when according to the second moment in the server cluster system are handed over
The configuration of the server of easy data, determines the transaction data that second moment corresponding server cluster system can be handled
Amount;If the transaction data amount that second moment corresponding server cluster system can be handled is corresponding greater than the object time
Transaction data amount, then increase before the object time in the server cluster system handle transaction data server
Quantity and/or improve the configuration that at least one server of transaction data is handled in the server cluster system;If described
The transaction data amount that second moment corresponding server cluster system can be handled is less than the corresponding number of deals of the object time
According to amount, then reduces the quantity for handling the server of transaction data in the server cluster system and/or reduce the server
The configuration of at least one server of transaction data is handled in group system.
The third aspect, the embodiment of the present invention also provides a kind of computer readable storage medium, including instruction, when it is being calculated
When being run on machine so that computer execute as above-mentioned first aspect or first aspect arbitrarily as described in server cluster system pipe
Reason method.
Fourth aspect, the embodiment of the invention also provides a kind of computer program products, when run on a computer,
So that computer execute as above-mentioned first aspect or first aspect arbitrarily as described in server cluster system management method.
The aspects of the invention (i.e. first aspect~fourth aspect) or other aspects are in the following description
Meeting more straightforward.
Detailed description of the invention
To describe the technical solutions in the embodiments of the present invention more clearly, make required in being described below to embodiment
Attached drawing is briefly introduced, it should be apparent that, drawings in the following description are only some embodiments of the invention, for this
For the those of ordinary skill in field, without any creative labor, it can also be obtained according to these attached drawings
His attached drawing.
Fig. 1 a is a kind of system architecture schematic diagram of server cluster system provided in an embodiment of the present invention;
Fig. 1 b is the system architecture schematic diagram of another server cluster system provided in an embodiment of the present invention;
Fig. 2 is a kind of corresponding overall flow figure of management method of server cluster system provided in an embodiment of the present invention;
Fig. 3 is a kind of structural schematic diagram of the managing device of server cluster system provided in an embodiment of the present invention.
Specific embodiment
To make the objectives, technical solutions, and advantages of the present invention clearer, below in conjunction with attached drawing to the present invention make into
It is described in detail to one step, it is clear that described embodiments are only a part of the embodiments of the present invention, rather than whole implementation
Example.Based on the embodiments of the present invention, obtained by those of ordinary skill in the art without making creative efforts
All other embodiment, shall fall within the protection scope of the present invention.
Fig. 1 a is a kind of system architecture schematic diagram of server cluster system provided in an embodiment of the present invention, such as Fig. 1 a institute
Show, may include at least one server, such as server 101, server 102,103 and of server in server cluster system
Server 104.Wherein, it can connect between any two server at least one server, such as can be by wired
Mode realizes connection, or can also realize connection wirelessly, is specifically not construed as limiting.
It, can be according to the first default distribution if server cluster system receives trading processing request in specific implementation
Trading processing is requested corresponding transaction allocation to any one server in 101~server of server 104 or appointed by mechanism
It anticipates multiple servers, so that trading processing requests corresponding transaction data can be handled by server cluster system.Wherein,
The distribution mechanism that one default distribution mechanism can be rule of thumb arranged for those skilled in the art, or may be user according to
The distribution mechanism of actual conditions setting, is specifically not construed as limiting.In one example, table 1 is one kind provided in an embodiment of the present invention
The schematic table of first default distribution mechanism.
A kind of table 1: signal of first default distribution mechanism
Transaction data amount (M) | Destination server |
(0,50] | Server 101 |
(50,120] | Server 101, server 103 |
(120,200] | Server 102, server 103 |
(200,500] | Server 102, server 103, server 104 |
(500,700] | 101~server of server 104 |
It as shown in table 1, may include the corresponding relationship of transaction data amount and destination server in the first default distribution mechanism,
If it is determined that trading processing requests corresponding transaction data amount to be less than or equal to 50M, then the transaction data can be distributed to service
Device 101 is handled;If it is determined that trading processing requests corresponding transaction data amount to be greater than 50M and is less than or equal to 120M, then may be used
The transaction data is distributed to server 101 and server 103 is handled;If it is determined that trading processing requests corresponding transaction
Data volume be greater than 120M and be less than or equal to 200M, then the transaction data can be distributed to server 102 and server 103 into
Row processing;It, then can should if it is determined that trading processing requests corresponding transaction data amount to be greater than 200M and is less than or equal to 500M
Transaction data is distributed to server 102, server 103 and server 104 and is handled;If it is determined that trading processing request is corresponding
Transaction data amount is greater than 500M and is less than or equal to 700M, then the transaction data can be distributed to 101~server of server
104 are handled.Correspondingly, if destination server include it is multiple, transaction data can be completed in multiple destination servers
After processing, the corresponding transaction data processing result of multiple destination servers is obtained, merging obtains server cluster system pair
The processing result for the transaction data answered.
Further, by taking server 101 handles the first transaction data as an example, if handling the first number of deals in server 101
During, since server 101 breaks down and it is processed to cause the first transaction data that can not continue, at this point it is possible to press
According to the second default distribution mechanism by the first transaction data distribute in 102~server of server 104 any one or it is any
Multiple servers.Wherein, the second default distribution mechanism can be the distribution mechanism that those skilled in the art are rule of thumb arranged, or
Person may be the distribution mechanism that user is arranged according to the actual situation, specifically be not construed as limiting.In the embodiment of the present invention, second is default
The implementation procedure of distribution mechanism can be realized that details are not described herein again according to the implementation procedure of the first distribution mechanism.
It should be noted that the first default distribution mechanism and the second default distribution mechanism that Fig. 1 a is illustrated can be and pass through
It is artificial to realize distribution, for example, artificial mode can be passed through if server cluster system receives trading processing request
Corresponding transaction data is requested to distribute to corresponding destination server trading processing.
Fig. 1 b is the system architecture schematic diagram of another server cluster system provided in an embodiment of the present invention, such as Fig. 1 b institute
Show, may include that (equipment 105 anticipated out as shown in Figure 1 b, is set equipment 106 at least one server in server cluster system
Standby 107 and equipment 108) and management at least one server central server 110.Wherein, central server 110 can with extremely
Each server connection in a few server, to realize the communication with each server.
In specific implementation, default distribution mechanism (such as the first default distribution mechanism can store in central server 110
With the second default distribution mechanism), by taking the communication process of central server 110 and server 105 as an example, in one example, in
Central server 110 can be connect by wired mode (such as cable, optical fiber) with server 105, in this way, central server 110
If receiving trading processing request, corresponding transaction data amount and default distribution mechanism can be requested to determine according to trading processing
Destination server (such as server 105), and then trading processing can be requested by corresponding transaction data by cable or optical fiber
It is sent to server 105, and server 105 can be received by cable or optical fiber and handled at the transaction data that transaction data obtains
Manage result.Correspondingly, if central server 110 detects that server 105 breaks down during handling transaction data,
The corresponding standby server of server 105 (such as server 106 and server 107) can be determined according to default distribution mechanism,
Then the currently processed transaction data of server 105 can be sent to server 106 and server 107 by central server 110.?
In another example, central server 110 can (such as microwave communication, satellite communication) and server 105 wirelessly
Connection, in this way, trading processing can be requested corresponding transaction data by sending the signal of predeterminated frequency by central server 110
It is transferred to server 105, and can receive the letter of the predeterminated frequency including transaction data processing result of the transmission of server 105
Number.
Based on the system architecture that Fig. 1 a and Fig. 1 b are illustrated, Fig. 2 is a kind of server cluster provided in an embodiment of the present invention
The corresponding flow diagram of the management method of system, this method comprises:
Step 201, N number of moment of multiple servers before the first moment respectively corresponds in acquisition server cluster system
Transaction data amount.
In the embodiment of the present invention, N can be rule of thumb configured by those skilled in the art, or can also be according to reality
Determination is tested, is specifically not construed as limiting.Management method in the embodiment of the present invention is described so that N is 100 as an example below.
In specific implementation, if server cluster system is server cluster system shown in Fig. 1 a, available service
The transaction data amount of 100 moment time-division other places reason of each server before the first moment in 101~server of device 103,
And it can be directed to each moment in 100 moment, 101~server of server 103 is handled total when counting each moment
Transaction data amount, obtain corresponding transaction data amount of each moment.Correspondingly, if server cluster system is shown in Fig. 1 b
Then presetting database has can be set in central server 110 in server cluster system;In one example, presetting database
In total transaction data amount of server cluster system processing therefore can be pre- by obtaining when can store each moment
If the data stored in database, each moment corresponding transaction data amount in 100 moment is determined;In another example,
Each server be can store in 105~server of server 108 in presetting database in 100 moment at each moment
The transaction data amount of reason can determine that each moment is corresponding in 100 moment by obtaining the data stored in presetting database
Transaction data amount.
Step 202, it predicts to obtain the corresponding transaction data of object time according to N number of moment corresponding transaction data amount
Amount.
Herein, object time can be the first moment, or may be the first moment after sometime, specifically not
It limits.
In one possible implementation, the corresponding number of deals of object time can be predicted by way of machine learning
According to amount.Specifically, available first initial model, and can be by the corresponding friendship of got in step 201 100 moment
Easy training data of the data volume as the first initial model, the first initial model of training, obtains the first model;Further, may be used
To use the first model prediction to obtain the corresponding first transaction data amount of object time.In the embodiment of the present invention, the first introductory die
Type can be multinomial model, and wherein the number of multinomial model can be set according to the actual situation by those skilled in the art
It sets, for example can be cubic polynomial model, perhaps or quartic polynomial model or can also be quintic algebra curve
Model is specifically not construed as limiting.
By taking the first initial model is quartic polynomial model as an example, then the corresponding quartic polynomial structure of the first initial model
It can satisfy following expression:
Wherein, tiIt can be moment, f1(ti) can be using the t of the first model predictioniMoment corresponding transaction data amount,
A, b, c, d and e can be the unknown parameter in quartic polynomial model.
It, can be initial as first using 100 moment and 100 moment corresponding transaction data amount in specific implementation
The input of model is fitted the corresponding value of parameter a, b, c, d and e determined in the first initial model using least square method.
For example, if 100 moment be respectively the moment 1, the moment 2, the moment 3 ..., moment 98, moment 99, moment 100, and at 100
Carving corresponding transaction data amount is x1、x2、x3、……、x98、x99、x100, then can be by using the side of least square method
The following equation of formula solution determines the corresponding value of parameter a, b, c, d and e:
Correspondingly, however, it is determined that the corresponding value of parameter a, b, c, d and e is respectively 5.7,2.6,1.2,4.3 and 8, then and the
One model can be with are as follows:
Wherein, the first mould can be made using the parameter value 5.7,2.6,1.2,4.3 and 8 that least square method fitting is determined
The prediction effect of type is preferable, that is, when using parameter value 5.7,2.6,1.2,4.3 and 8, obtained using the first model prediction 100
Moment corresponding transaction data amount real trade data volume corresponding with 100 moment is closer to.
It is possible to further use the corresponding transaction data amount of the first model prediction object time, for example, if object time
For moment tn, then t at the time of being obtained using the first model predictionnCorresponding transaction data amount can be with are as follows:
In the embodiment of the present invention, by being predicted based on multinomial model the corresponding transaction data amount of object time,
The transaction data amount that prediction can be made to obtain is more accurate;And it is more by using the corresponding transaction data amount adjustment of N number of moment
Parameter in item formula model, the parameter for the multinomial model that training can be made to obtain is more accurate, i.e., multinomial model is pre-
Survey effect is preferable, so as to further improve the accuracy for the transaction data amount that prediction obtains.
In the embodiment of the present invention, t at the time of being obtained according to the first model predictionnCorresponding transaction data amount determines moment tn
The mode of corresponding transaction data amount can there are many, in an example (for ease of description, referred to as example one), Ke Yizhi
Meet t at the time of obtaining the first model predictionnCorresponding transaction data amount is determined as moment tnCorresponding transaction data amount.Another
In one example (for ease of description, referred to as example two), t at the time of the first model prediction being obtainednCorresponding transaction
Data volume is as moment tnThe first part of corresponding transaction data amount, and can using other model predictions obtain at the time of tn
Corresponding transaction data amount is as moment tnThe second part of corresponding transaction data amount, obtains according to first part and second part
To moment tnCorresponding transaction data amount.
In example one, due to t at the time of the first model prediction obtainsnCorresponding transaction data amount is f1(tn) (such as
10000) t at the time of, therefore, determination obtains in step 202nCorresponding transaction data amount can be 10000.
In example two, it can determine that 100 moment corresponding transaction data is to belong to periodical number using default means
According to still falling within aperiodicity data, wherein default means can be rule of thumb configured by those skilled in the art, or
It can also be configured, specifically be not construed as limiting according to actual needs.In one possible implementation, it may be predetermined that clothes
The execution period of device group system of being engaged in processing current business obtains the corresponding transaction data of processing current business in the execution period
Amount, and then determine that 100 moment corresponding transaction data belongs to periodicity according to the transaction data amount of a cycle got
Data still fall within aperiodicity data.For example, if 100 moment corresponding transaction data is to generate the corresponding friendship of order
Easy data volume, and server cluster system A executes that generate the transaction cycle of order be 1440 minutes, then available server set
Group system A executes the corresponding transaction data of exchange for generating order in 1440 minutes before the first moment;If 1440 points
Transaction data in clock is periodic data, then can determine that 100 moment corresponding transaction data belongs to periodic data, if
Transaction data in 1440 minutes is aperiodicity data, then can determine that 100 moment corresponding transaction data belongs to non-week
Phase property data.
Separately below from the specific implementation process of situation one and the description prediction transaction data of situation two, situation one is 100
Moment corresponding transaction data amount predicts that the specific implementation process of transaction data, situation two are 100 when belonging to aperiodicity data
A moment corresponding transaction data amount predicts the specific implementation process of transaction data when belonging to periodic data.
Situation one
If it is determined that 100 moment corresponding transaction data amount belongs to aperiodicity data, then available second introductory die
Type, and the second initial model of 100 moment corresponding transaction data amount training can be used, obtain the second model.Further,
The second model prediction can be used and obtain moment tnCorresponding transaction data amount, and can be according to using the first model prediction to obtain
At the time of tnCorresponding transaction data amount and moment t is obtained using the second model predictionnCorresponding transaction data amount, determines the moment
tnCorresponding transaction data amount.
Wherein, the second initial model can be second multinomial exponential smoothing model, the corresponding secondary index knot of the second initial model
Structure can satisfy following expression:
Wherein, tiIt can be moment, s (ti) it can be tiMoment corresponding slow growth variable, x (ti) it can be moment ti
When the transaction data amount that handles, α, β can be the smoothing factor that is arranged in second multinomial exponential smoothing model, and the value of α, β can be by these
Field technical staff is rule of thumb configured, for example α can be set to 0.7, β can be set to 0.8, is specifically not construed as limiting.
In specific implementation, by taking moment 1, moment 2 and moment 3 as an example, the specific implementation procedure of above-mentioned expression formula can be with are as follows:
Step q1, by moment 1 and the corresponding transaction data amount x of moment 11Substitute into the corresponding expression of second multinomial exponential smoothing model
Formula, the available corresponding equation of moment 1:
Wherein, s0And r0Value can be rule of thumb configured by those skilled in the art, for example s can be set0And r0
It is 0.By solving the corresponding equation of moment 1, the corresponding s of moment 1 can be determined1And r1。
Step q2, by the corresponding transaction data amount x of moment 22With step q1Middle determination obtains the corresponding s of moment 11、r1It substitutes into
The corresponding expression formula of second multinomial exponential smoothing model, the available corresponding equation of moment 2:
By solving the corresponding equation of moment 2, the corresponding s of moment 2 can be determined2And r2。
Step q3, by the corresponding transaction data amount x of moment 33With step q2Middle determination obtains the corresponding s of moment 22And r2It substitutes into
The corresponding expression formula of second multinomial exponential smoothing model, the available corresponding equation of moment 3:
By solving the corresponding equation of moment 3, the corresponding s of moment 3 can be determined3And r3。
According to step q1~step q3It is found that by successively by the moment 1, the moment 2, the moment 3 ..., moment 98, moment 99
The corresponding expression formula of second multinomial exponential smoothing model is substituted into, can finally determine the corresponding s of moment 100100And r100。
It is possible to further according to the corresponding s of moment 100100And r100At the time of determination is obtained using the second model prediction
tnCorresponding transaction data amount;In one possible implementation, t at the time of being obtained using the second model predictionnCorresponding friendship
Easy data volume can be with are as follows:
f2(tn)=s100+(tn-t100)r100
Wherein, t100It can be the end time in 100 moment, i.e. moment 100.
For example, if the 100th moment in 100 moment is the moment 100, object time is the moment 101, then uses
101 corresponding transaction data amounts can be at the time of second model prediction obtains are as follows:
f2(101)=s100+(101-100)r100
If object time is the moment 110,110 corresponding transaction data amounts can at the time of being obtained using the second model prediction
With are as follows:
f2(110)=s100+(110-100)r100
Further, in one possible implementation, according to t at the time of using the first model prediction to obtainnIt is corresponding
Transaction data amount f1(tn) and t at the time of obtained using the second model predictionnCorresponding transaction data amount f2(tn), determination obtains
At the time of tnCorresponding transaction data amount f (tn) can be with are as follows:
f(tn)=τ1f1(tn)+ξ1f2(tn)
Wherein, τ1It can be the corresponding weight of the first model, ξ1It can be the corresponding weight of the second model, the first model pair
The weight and the corresponding weight of the second model answered can be rule of thumb configured by those skilled in the art, for example can be set
The corresponding weight of first model and the corresponding weight of the second model and be 1, be specifically not construed as limiting.
It should be noted that above-mentioned carried out so that object time is a certain moment (such as moment 101 or moment 110) as an example
Description, in other possible embodiments, object time also may include multiple (two or more) moment, for example,
Object time includes moment 101, moment 102 and moment 103.
Situation two
If it is determined that 100 moment corresponding transaction data amount belongs to periodic data, then available third initial model,
And 100 moment corresponding transaction data amount training third initial model can be used, and obtain third model further, it can
To use third model prediction to obtain moment tnCorresponding transaction data amount, and can be according to using the first model prediction to obtain
Moment tnCorresponding transaction data amount and moment t is obtained using third model predictionnCorresponding transaction data amount, determines moment tn
Corresponding transaction data amount.
Wherein, third initial model can be Three-exponential Smoothing model, the corresponding index knot three times of third initial model
Structure can satisfy following expression:
Wherein, tiIt can be the moment,It can be tiThe corresponding period at moment increases variable,It can be tiMoment is corresponding
Slow growth variable,It can be tiThe corresponding period at moment slowly increases variable, and γ, μ and σ can be Three-exponential Smoothing
The value of the smoothing factor being arranged in model, γ, μ and σ can be rule of thumb configured by those skilled in the art, for example γ can
0.8, σ, which be can be set to, to be set as 0.7, μ can be set to 0.8, specifically be not construed as limiting.
In specific implementation, by successively by the moment 1, the moment 2, the moment 3 ..., moment 98, moment 99 substitute into index three times
The corresponding expression formula of smoothing model can finally determine the corresponding y of moment 100100、w100And l100.It should be noted that determining
The corresponding y of moment 100100、w100And l100Process be referred in second multinomial exponential smoothing model determine the corresponding s of moment 100100
And r100Process realized that details are not described herein again.
It is possible to further according to the corresponding y of moment 100100、w100And l100It is determining that third model prediction is used to obtain
Moment tnCorresponding transaction data amount;In one possible implementation, t at the time of being obtained using third model predictionnIt is corresponding
Transaction data amount can be with are as follows:
Wherein, t100It can be the end time in 100 moment, i.e. moment 100.
For example, if the 100th moment in 100 moment is the moment 100, object time is the moment 101, then uses
101 corresponding transaction data amounts can be at the time of second model prediction obtains are as follows:
f3(101)=y100+(101-100)w100+l1+1+(1-1)mod100
If object time is the moment 110,110 corresponding transaction data amounts can at the time of being obtained using the second model prediction
With are as follows:
f3(110)=y100+(110-100)w100+l110-100+1+(110-100-1)mod100
Further, in one possible implementation, according to t at the time of using the first model prediction to obtainnIt is corresponding
Transaction data amount f1(tn) and t at the time of obtained using third model predictionnCorresponding transaction data amount f3(tn), determination obtains
At the time of tnCorresponding transaction data amount f (tn) can be with are as follows:
f(tn)=τ2f1(tn)+ξ2f3(tn)
Wherein, τ2It can be the corresponding weight of the first model, ξ2It can be the corresponding weight of the second model, the first model pair
The weight and the corresponding weight of the second model answered can be rule of thumb configured by those skilled in the art, for example can be set
The corresponding weight of first model and the corresponding weight of the second model and be 1, be specifically not construed as limiting.
It should be noted that the first model corresponding weight and first model in situation one exist in the embodiment of the present invention
Corresponding weight can be identical in situation two, or can also be different, correspondingly, the second model corresponding weight in situation one
Corresponding weight can be identical in situation two with third model, or can also be different, and is specifically not construed as limiting.
In the embodiment of the present invention, by determining whether transaction data amount is periodic data, can choose different models
(i.e. the second model and third model) predicts periodic transaction data amount and acyclic transaction data amount respectively,
So that determining that the obtained corresponding transaction data amount of object time is more accurate, and then can be based on the transaction
The server cluster system of data volume management is more in line with actual conditions;And object time pair determined by the embodiment of the present invention
The transaction data amount answered is obtained by two model predictions, Individual forecast knot of the prediction result compared to two models
It is more accurate for fruit, so as to more improve ground management server group system based on the transaction data amount.
Step 203, according to the corresponding transaction data amount of the object time, the clothes are adjusted before the object time
The quantity of the server of transaction data is handled in device group system of being engaged in and/or handles the configuration of the server of transaction data.
By taking object time is the moment 102 as an example, in one possible implementation, it can be determined from 100 moment
Out apart from the moment 102 it is nearest at the time of (i.e. moment 100), and when the available moment 100 in server cluster system handle hand over
The configuration of the server of the quantity and processing transaction data of the server of easy data.It is taken when it is possible to further according to moment 100
The configuration that the quantity of the server of transaction data and the server of processing transaction data are handled in device group system of being engaged in, determines the moment
The transaction data amount (for example being 20000M) that 100 corresponding server cluster systems can be handled.
If 102 corresponding transaction data amounts are 10000M, explanation at the time of being obtained according to step 201~step 202 prediction
The processing capacity of server cluster system is inadequate in dealing with the transaction data of 20000M when moment 100, at this point it is possible at the moment
(such as moment 101) increases the quantity that the server of transaction data is handled in server cluster system before 102, and/or, it improves
In server cluster system handle transaction data at least one server hardware configuration (such as increase memory, upgraded version
Deng);Correspondingly, it if 102 corresponding transaction data amounts are 30000M at the time of being obtained according to step 201~step 202 prediction, says
The processing capacity of server cluster system is enough to handle the transaction data of 20000M when bright moment 100, at this point it is possible at the moment
(such as moment 101) reduces the quantity that the server of transaction data is handled in server cluster system before 102, and/or, it reduces
In server cluster system handle transaction data at least one server hardware configuration (such as reduce memory, reduce version
Deng).
It should be noted that if above-mentioned steps 202 predict obtained object time include multiple moment (such as the moment 101,
Moment 102 and moment 103), then it is above-mentioned described " to adjust in the server cluster system before the object time
Handle the quantity of the server of transaction data and/or handle the configuration of the server of transaction data " it is to be understood that described more
The service that transaction data is handled in the server cluster system is adjusted at the time of in a moment earliest before (such as moment 101)
The configuration of the server of the quantity and/or processing transaction data of device.
In the above embodiment of the present invention, N of multiple servers before the first moment in server cluster system is obtained
A moment corresponding transaction data amount, and according to corresponding transaction data amount of N number of moment predict to obtain object time corresponding
Transaction data amount, wherein at the time of object time is after the first moment or the first moment;When further, according to target
Corresponding transaction data amount is carved, the number for handling the server of transaction data in server cluster system is adjusted before object time
The configuration of the server of amount and/or processing transaction data.In the embodiment of the present invention, by using the number of deals before object time
According to the transaction data amount of amount prediction object time, server cluster system can be made to can be completed before object time arrival
Management to server cluster system, so as to improve the flexibility of management;And by advance to server cluster system into
The mode of row management, the server cluster system after can making management can be handled in time object time in object time and correspond to
Transaction data, so as to improve server cluster system processing transaction data efficiency.
For above method process, the embodiment of the present invention also provides a kind of managing device of server cluster system, the dress
The particular content set is referred to above method implementation.
Fig. 3 is a kind of structural schematic diagram of the managing device of server cluster system provided in an embodiment of the present invention, the dress
It sets and includes:
Module 301 is obtained, for obtaining N number of moment of multiple servers before the first moment in server cluster system
Corresponding transaction data amount;
Prediction module 302, it is corresponding for predicting to obtain object time according to N number of moment corresponding transaction data amount
Transaction data amount;At the time of the object time is after first moment or first moment;
Management module 303, for being adjusted before the object time according to the corresponding transaction data amount of the object time
The quantity of the server of transaction data is handled in the whole server cluster system and/or handles matching for the server of transaction data
It sets.
Optionally, the prediction module 302 is used for:
The first initial model is obtained, first initial model is multinomial model;
Using N number of moment corresponding transaction data amount training first initial model, the first model is obtained;
The corresponding first transaction data amount of the object time is obtained using first model prediction, and according at least to institute
It states the first transaction data amount and determines the corresponding transaction data amount of the object time.
Optionally, the prediction module 302 is used for:
If it is determined that N number of moment corresponding transaction data amount belongs to aperiodicity data, then the second initial model is obtained,
Using N number of moment corresponding transaction data amount training second initial model, the second model is obtained;Use described second
Model prediction obtains the corresponding second transaction data amount of the object time, and according to the first transaction data amount and described
Two transaction data amounts, determine the corresponding transaction data amount of the object time, and second initial model is double smoothing
Model;Alternatively,
If it is determined that N number of moment corresponding transaction data amount belongs to periodic data, then third initial model is obtained, is made
With N number of moment corresponding transaction data amount training third initial model, third model is obtained;Use the third mould
Type is predicted to obtain the corresponding third transaction data amount of the object time, is handed over according to the first transaction data amount and the third
Easy data volume, determines the corresponding transaction data amount of the object time, and the third initial model is Three-exponential Smoothing model.
Optionally, in N number of moment with first moment it is close at the time of be the second moment;
The prediction module 302 is used for:
Use the N-1 moment corresponding transaction data amount training institute in N number of moment in addition to second moment
The second initial model is stated, second moment corresponding second model is obtained;According to second moment corresponding second model
Transaction data amount corresponding with second moment trains second moment corresponding second model, obtains second mould
Type;Alternatively,
Use the N-1 moment corresponding transaction data amount training institute in N number of moment in addition to second moment
Third initial model is stated, second moment corresponding third model is obtained;According to second moment corresponding third model
Transaction data amount corresponding with second moment trains second moment corresponding third model, obtains the third mould
Type.
Optionally, the prediction module 302 is used for:
Obtain corresponding first weight of first model and the corresponding weight of second model;It is handed over according to described first
Easy data volume, corresponding first weight of first model, the second transaction data amount and the corresponding power of second model
Weight, determines the corresponding transaction data amount of the object time;Alternatively,
Obtain corresponding second weight of first model and the corresponding weight of the third model;It is handed over according to described first
Easy data volume, corresponding second weight of first model, the third transaction data amount and the corresponding power of the third model
Weight, determines the corresponding transaction data amount of the object time.
Optionally, the management module 303 is used for:
It obtains the quantity for handling the server of transaction data when the second moment in the server cluster system and processing is handed over
The configuration of the server of easy data;At the time of second moment is nearest before first moment;
The quantity and processing for handling the server of transaction data when according to the second moment in the server cluster system are handed over
The configuration of the server of easy data, determines the transaction data that second moment corresponding server cluster system can be handled
Amount;
If the transaction data amount that second moment corresponding server cluster system can be handled is greater than the target
Corresponding transaction data amount is carved, then increases before the object time and handles transaction data in the server cluster system
The quantity of server and/or the configuration for improving at least one server of processing transaction data in the server cluster system;
If it is corresponding that the transaction data amount that second moment corresponding server cluster system can be handled is less than the object time
Transaction data amount is then reduced described in the quantity and/or reduction for handling the server of transaction data in the server cluster system
The configuration of at least one server of transaction data is handled in server cluster system.
It can be seen from the above: in the above embodiment of the present invention, obtaining multiple services in server cluster system
N number of moment of the device before the first moment corresponding transaction data amount, and it is pre- according to corresponding transaction data amount of N number of moment
Measure the corresponding transaction data amount of object time, wherein at the time of object time is after the first moment or the first moment;Into
One step, according to the corresponding transaction data amount of object time, adjusts to handle in server cluster system before object time and hand over
The quantity of the server of easy data and/or the configuration for the server for handling transaction data.In the embodiment of the present invention, by using mesh
The transaction data amount for marking the transaction data amount prediction object time before the moment, can make server cluster system in target
It is carved into the management that can be completed before coming to server cluster system, so as to improve the flexibility of management;And by preparatory
To the mode that server cluster system is managed, server cluster system after management can be made object time can and
When processing target moment corresponding transaction data, so as to improve server cluster system processing transaction data efficiency.
Based on the same inventive concept, the embodiment of the invention also provides a kind of computer readable storage mediums, including instruct,
When run on a computer, so that computer executes the manager of server cluster system shown in above-mentioned Fig. 2 embodiment
Method.
Based on the same inventive concept, the embodiment of the invention also provides a kind of computer program products, when it is in computer
When upper operation, so that computer executes the management method of server cluster system shown in above-mentioned Fig. 2 embodiment.
It should be understood by those skilled in the art that, the embodiment of the present invention can provide as method or computer program product.
Therefore, complete hardware embodiment, complete software embodiment or embodiment combining software and hardware aspects can be used in the present invention
Form.It is deposited moreover, the present invention can be used to can be used in the computer that one or more wherein includes computer usable program code
The shape for the computer program product implemented on storage media (including but not limited to magnetic disk storage, CD-ROM, optical memory etc.)
Formula.
The present invention be referring to according to the method for the embodiment of the present invention, the process of equipment (system) and computer program product
Figure and/or block diagram describe.It should be understood that every one stream in flowchart and/or the block diagram can be realized by computer program instructions
The combination of process and/or box in journey and/or box and flowchart and/or the block diagram.It can provide these computer programs
Instruct the processor of general purpose computer, special purpose computer, Embedded Processor or other programmable data processing devices to produce
A raw machine, so that being generated by the instruction that computer or the processor of other programmable data processing devices execute for real
The device for the function of being specified in present one or more flows of the flowchart and/or one or more blocks of the block diagram.
These computer program instructions, which may also be stored in, is able to guide computer or other programmable data processing devices with spy
Determine in the computer-readable memory that mode works, so that it includes referring to that instruction stored in the computer readable memory, which generates,
Enable the manufacture of device, the command device realize in one box of one or more flows of the flowchart and/or block diagram or
The function of being specified in multiple boxes.
These computer program instructions also can be loaded onto a computer or other programmable data processing device, so that counting
Series of operation steps are executed on calculation machine or other programmable devices to generate computer implemented processing, thus in computer or
The instruction executed on other programmable devices is provided for realizing in one or more flows of the flowchart and/or block diagram one
The step of function of being specified in a box or multiple boxes.
Although preferred embodiments of the present invention have been described, it is created once a person skilled in the art knows basic
Property concept, then additional changes and modifications may be made to these embodiments.So it includes excellent that the following claims are intended to be interpreted as
It selects embodiment and falls into all change and modification of the scope of the invention.
Obviously, various changes and modifications can be made to the invention without departing from essence of the invention by those skilled in the art
Mind and range.In this way, if these modifications and changes of the present invention belongs to the range of the claims in the present invention and its equivalent technologies
Within, then the present invention is also intended to include these modifications and variations.
Claims (14)
1. a kind of management method of server cluster system, which is characterized in that the described method includes:
Obtain N number of moment corresponding transaction data of multiple servers before the first moment in server cluster system
Amount;
It predicts to obtain the corresponding transaction data amount of object time according to N number of moment corresponding transaction data amount;The target
At the time of moment is after first moment or first moment;
According to the corresponding transaction data amount of the object time, the server cluster system is adjusted before the object time
The quantity of the server of middle processing transaction data and/or the configuration for the server for handling transaction data.
2. the method according to claim 1, wherein described according to N number of moment corresponding transaction data amount
Prediction obtains the corresponding transaction data amount of object time, comprising:
The first initial model is obtained, first initial model is multinomial model;
Using N number of moment corresponding transaction data amount training first initial model, the first model is obtained;
The corresponding first transaction data amount of the object time is obtained using first model prediction, and according at least to described
One transaction data amount determines the corresponding transaction data amount of the object time.
3. according to the method described in claim 2, it is characterized in that, described determine institute according at least to the first transaction data amount
State the corresponding transaction data amount of object time, comprising:
If it is determined that N number of moment corresponding transaction data amount belongs to aperiodicity data, then the second initial model is obtained, used
The corresponding transaction data amount training of N number of moment second initial model, obtains the second model;Use second model
Prediction obtains the corresponding second transaction data amount of the object time, and is handed over according to the first transaction data amount and described second
Easy data volume, determines the corresponding transaction data amount of the object time, and second initial model is second multinomial exponential smoothing model;
Alternatively,
If it is determined that N number of moment corresponding transaction data amount belongs to periodic data, then third initial model is obtained, institute is used
Corresponding transaction data amount training of the N number of moment third initial model is stated, third model is obtained;It is pre- using the third model
The corresponding third transaction data amount of the object time is measured, according to the first transaction data amount and the third number of deals
According to amount, the corresponding transaction data amount of the object time is determined, the third initial model is Three-exponential Smoothing model.
4. according to the method described in claim 3, it is characterized in that, in N number of moment with first moment it is close when
Carving was the second moment;
It is described to use corresponding transaction data amount training of N number of moment second initial model, the second model is obtained, is wrapped
It includes:
Use N-1 moment in N number of moment in addition to second moment corresponding transaction data amount training described the
Two initial models obtain second moment corresponding second model;According to second moment corresponding second model and institute
Corresponding second model of corresponding transaction data amount training second moment at the second moment is stated, second model is obtained;Or
Person,
It is described to use corresponding transaction data amount training of the N number of moment third initial model, third model is obtained, is wrapped
It includes:
Use N-1 moment in N number of moment in addition to second moment corresponding transaction data amount training described the
Three initial models obtain second moment corresponding third model;According to second moment corresponding third model and institute
Corresponding third model of corresponding transaction data amount training second moment at the second moment is stated, the third model is obtained.
5. according to the method described in claim 3, it is characterized in that, described according to the first transaction data amount and described second
Transaction data amount determines the corresponding transaction data amount of the object time, comprising:
Obtain corresponding first weight of first model and the corresponding weight of second model;According to first number of deals
According to amount, corresponding first weight of first model, the second transaction data amount and the corresponding weight of second model, really
Determine the corresponding transaction data amount of the object time;
It is described according to the first transaction data amount and the third transaction data amount, determine the corresponding transaction of the object time
Data volume, comprising:
Obtain corresponding second weight of first model and the corresponding weight of the third model;According to first number of deals
According to amount, corresponding second weight of first model, the third transaction data amount and the corresponding weight of the third model, really
Determine the corresponding transaction data amount of the object time.
6. the method according to any one of claims 1 to 5, which is characterized in that described corresponding according to the object time
Transaction data amount, the server that transaction data is handled in the server cluster system is adjusted before the object time
The configuration of the server of quantity and/or processing transaction data, comprising:
Obtain the quantity for handling the server of transaction data when the second moment in the server cluster system and processing number of deals
According to server configuration;At the time of second moment is nearest before first moment;
The quantity and processing number of deals of the server of transaction data are handled when according to the second moment in the server cluster system
According to server configuration, determine the transaction data amount that second moment corresponding server cluster system can be handled;
If the transaction data amount that second moment corresponding server cluster system can be handled is greater than the object time pair
The transaction data amount answered then increases the service that transaction data is handled in the server cluster system before the object time
The quantity of device and/or the configuration for improving at least one server of processing transaction data in the server cluster system;If institute
It states the transaction data amount that the second moment corresponding server cluster system can be handled and is less than the corresponding transaction of the object time
Data volume then reduces the quantity for handling the server of transaction data in the server cluster system and/or reduces the service
The configuration of at least one server of transaction data is handled in device group system.
7. a kind of managing device of server cluster system, which is characterized in that described device includes:
Module is obtained, it is right respectively for obtaining N number of moment of multiple servers before the first moment in server cluster system
The transaction data amount answered;
Prediction module, for predicting to obtain the corresponding number of deals of object time according to N number of moment corresponding transaction data amount
According to amount;At the time of the object time is after first moment or first moment;
Management module, described in being adjusted before the object time according to the corresponding transaction data amount of the object time
The quantity of the server of transaction data is handled in server cluster system and/or handles the configuration of the server of transaction data.
8. device according to claim 7, which is characterized in that the prediction module is used for:
The first initial model is obtained, first initial model is multinomial model;
Using N number of moment corresponding transaction data amount training first initial model, the first model is obtained;
The corresponding first transaction data amount of the object time is obtained using first model prediction, and according at least to described
One transaction data amount determines the corresponding transaction data amount of the object time.
9. device according to claim 8, which is characterized in that the prediction module is used for:
If it is determined that N number of moment corresponding transaction data amount belongs to aperiodicity data, then the second initial model is obtained, used
The corresponding transaction data amount training of N number of moment second initial model, obtains the second model;Use second model
Prediction obtains the corresponding second transaction data amount of the object time, and is handed over according to the first transaction data amount and described second
Easy data volume, determines the corresponding transaction data amount of the object time, and second initial model is second multinomial exponential smoothing model;
Alternatively,
If it is determined that N number of moment corresponding transaction data amount belongs to periodic data, then third initial model is obtained, institute is used
Corresponding transaction data amount training of the N number of moment third initial model is stated, third model is obtained;It is pre- using the third model
The corresponding third transaction data amount of the object time is measured, according to the first transaction data amount and the third number of deals
According to amount, the corresponding transaction data amount of the object time is determined, the third initial model is Three-exponential Smoothing model.
10. device according to claim 9, which is characterized in that in N number of moment with first moment it is close when
Carving was the second moment;
The prediction module is used for:
Use N-1 moment in N number of moment in addition to second moment corresponding transaction data amount training described the
Two initial models obtain second moment corresponding second model;According to second moment corresponding second model and institute
Corresponding second model of corresponding transaction data amount training second moment at the second moment is stated, second model is obtained;Or
Person,
Use N-1 moment in N number of moment in addition to second moment corresponding transaction data amount training described the
Three initial models obtain second moment corresponding third model;According to second moment corresponding third model and institute
Corresponding third model of corresponding transaction data amount training second moment at the second moment is stated, the third model is obtained.
11. device according to claim 9, which is characterized in that the prediction module is used for:
Obtain corresponding first weight of first model and the corresponding weight of second model;According to first number of deals
According to amount, corresponding first weight of first model, the second transaction data amount and the corresponding weight of second model, really
Determine the corresponding transaction data amount of the object time;Alternatively,
Obtain corresponding second weight of first model and the corresponding weight of the third model;According to first number of deals
According to amount, corresponding second weight of first model, the third transaction data amount and the corresponding weight of the third model, really
Determine the corresponding transaction data amount of the object time.
12. device according to any one of claims 7 to 11, which is characterized in that the management module is used for:
Obtain the quantity for handling the server of transaction data when the second moment in the server cluster system and processing number of deals
According to server configuration;At the time of second moment is nearest before first moment;
The quantity and processing number of deals of the server of transaction data are handled when according to the second moment in the server cluster system
According to server configuration, determine the transaction data amount that second moment corresponding server cluster system can be handled;
If the transaction data amount that second moment corresponding server cluster system can be handled is greater than the object time pair
The transaction data amount answered then increases the service that transaction data is handled in the server cluster system before the object time
The quantity of device and/or the configuration for improving at least one server of processing transaction data in the server cluster system;If institute
It states the transaction data amount that the second moment corresponding server cluster system can be handled and is less than the corresponding transaction of the object time
Data volume then reduces the quantity for handling the server of transaction data in the server cluster system and/or reduces the service
The configuration of at least one server of transaction data is handled in device group system.
13. a kind of computer readable storage medium, which is characterized in that including instruction, when run on a computer, make to succeed in one's scheme
Calculation machine executes such as method as claimed in any one of claims 1 to 6.
14. a kind of computer program product, which is characterized in that when run on a computer, so that computer is executed as weighed
Benefit requires 1 to 6 described in any item methods.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201910286423.7A CN110109800A (en) | 2019-04-10 | 2019-04-10 | A kind of management method and device of server cluster system |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201910286423.7A CN110109800A (en) | 2019-04-10 | 2019-04-10 | A kind of management method and device of server cluster system |
Publications (1)
Publication Number | Publication Date |
---|---|
CN110109800A true CN110109800A (en) | 2019-08-09 |
Family
ID=67485282
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201910286423.7A Pending CN110109800A (en) | 2019-04-10 | 2019-04-10 | A kind of management method and device of server cluster system |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN110109800A (en) |
Cited By (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN111179076A (en) * | 2019-12-18 | 2020-05-19 | 平安证券股份有限公司 | IT system intelligent management method, IT system intelligent management device and computer readable storage medium |
CN111638962A (en) * | 2020-06-08 | 2020-09-08 | 杭州复杂美科技有限公司 | Dynamic expansion and contraction method, device and storage medium for transaction pool |
CN112036852A (en) * | 2020-08-25 | 2020-12-04 | 北京达佳互联信息技术有限公司 | Payment processing method and device, server and electronic equipment |
CN112764893A (en) * | 2019-11-04 | 2021-05-07 | 华为技术有限公司 | Data processing method and data processing system |
WO2021115054A1 (en) * | 2019-12-09 | 2021-06-17 | 聚好看科技股份有限公司 | Method for adjusting node configuration in cluster system, and server |
Citations (11)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN103490956A (en) * | 2013-09-22 | 2014-01-01 | 杭州华为数字技术有限公司 | Self-adaptive energy-saving control method, device and system based on traffic predication |
CN105912436A (en) * | 2015-09-17 | 2016-08-31 | 乐视网信息技术(北京)股份有限公司 | Method and device for predicting system computing resource based on exponential smoothing prediction |
CN106549772A (en) * | 2015-09-16 | 2017-03-29 | 华为技术有限公司 | Resource prediction method, system and capacity management device |
CN107146015A (en) * | 2017-05-02 | 2017-09-08 | 联想(北京)有限公司 | Multivariate Time Series Forecasting Methodology and system |
CN107527124A (en) * | 2017-10-13 | 2017-12-29 | 众安信息技术服务有限公司 | The method and apparatus for generating industry basic side combination forecasting |
CN108230049A (en) * | 2018-02-09 | 2018-06-29 | 新智数字科技有限公司 | The Forecasting Methodology and system of order |
CN108845881A (en) * | 2018-05-30 | 2018-11-20 | 有米科技股份有限公司 | The method and device of server capacity dynamic adjustment |
CN108900343A (en) * | 2018-07-04 | 2018-11-27 | 中国人民解放军国防科技大学 | Local storage-based resource prediction and scheduling method for cloud server |
CN108984304A (en) * | 2018-07-11 | 2018-12-11 | 广东亿迅科技有限公司 | Server expansion calculation method and device based on regression equation |
CN109034505A (en) * | 2018-09-14 | 2018-12-18 | 国网河北省电力有限公司电力科学研究院 | Metering table demand on-line prediction method and terminal device |
CN109389329A (en) * | 2018-11-13 | 2019-02-26 | 平安科技(深圳)有限公司 | Test performance prediction technique, device, medium and the electronic equipment of test macro |
-
2019
- 2019-04-10 CN CN201910286423.7A patent/CN110109800A/en active Pending
Patent Citations (11)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN103490956A (en) * | 2013-09-22 | 2014-01-01 | 杭州华为数字技术有限公司 | Self-adaptive energy-saving control method, device and system based on traffic predication |
CN106549772A (en) * | 2015-09-16 | 2017-03-29 | 华为技术有限公司 | Resource prediction method, system and capacity management device |
CN105912436A (en) * | 2015-09-17 | 2016-08-31 | 乐视网信息技术(北京)股份有限公司 | Method and device for predicting system computing resource based on exponential smoothing prediction |
CN107146015A (en) * | 2017-05-02 | 2017-09-08 | 联想(北京)有限公司 | Multivariate Time Series Forecasting Methodology and system |
CN107527124A (en) * | 2017-10-13 | 2017-12-29 | 众安信息技术服务有限公司 | The method and apparatus for generating industry basic side combination forecasting |
CN108230049A (en) * | 2018-02-09 | 2018-06-29 | 新智数字科技有限公司 | The Forecasting Methodology and system of order |
CN108845881A (en) * | 2018-05-30 | 2018-11-20 | 有米科技股份有限公司 | The method and device of server capacity dynamic adjustment |
CN108900343A (en) * | 2018-07-04 | 2018-11-27 | 中国人民解放军国防科技大学 | Local storage-based resource prediction and scheduling method for cloud server |
CN108984304A (en) * | 2018-07-11 | 2018-12-11 | 广东亿迅科技有限公司 | Server expansion calculation method and device based on regression equation |
CN109034505A (en) * | 2018-09-14 | 2018-12-18 | 国网河北省电力有限公司电力科学研究院 | Metering table demand on-line prediction method and terminal device |
CN109389329A (en) * | 2018-11-13 | 2019-02-26 | 平安科技(深圳)有限公司 | Test performance prediction technique, device, medium and the electronic equipment of test macro |
Non-Patent Citations (1)
Title |
---|
全国成人教育国际贸易专科系列教材编委会: "《综合辅导大纲及模拟试卷》", 中国对外经济贸易出版社 * |
Cited By (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN112764893A (en) * | 2019-11-04 | 2021-05-07 | 华为技术有限公司 | Data processing method and data processing system |
CN112764893B (en) * | 2019-11-04 | 2024-03-26 | 华为技术有限公司 | Data processing method and data processing system |
WO2021115054A1 (en) * | 2019-12-09 | 2021-06-17 | 聚好看科技股份有限公司 | Method for adjusting node configuration in cluster system, and server |
CN111179076A (en) * | 2019-12-18 | 2020-05-19 | 平安证券股份有限公司 | IT system intelligent management method, IT system intelligent management device and computer readable storage medium |
CN111638962A (en) * | 2020-06-08 | 2020-09-08 | 杭州复杂美科技有限公司 | Dynamic expansion and contraction method, device and storage medium for transaction pool |
CN112036852A (en) * | 2020-08-25 | 2020-12-04 | 北京达佳互联信息技术有限公司 | Payment processing method and device, server and electronic equipment |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN110109800A (en) | A kind of management method and device of server cluster system | |
CN107403296B (en) | Transport capacity configuration method and device | |
CN109412829A (en) | A kind of prediction technique and equipment of resource distribution | |
CN104243598B (en) | A kind of information recommendation method and device | |
CN106959894B (en) | Resource allocation method and device | |
CN107229518A (en) | A kind of distributed type assemblies training method and device | |
CN103345514A (en) | Streamed data processing method in big data environment | |
CN109447355A (en) | Dispatching optimization method, device, medium and the calculating equipment of articles from the storeroom | |
CN105657449B (en) | A kind of video code conversion distribution method, device and video code conversion system | |
CN109842563A (en) | Content delivery network flow dispatching method, device and computer readable storage medium | |
CN109885397A (en) | The loading commissions migration algorithm of time delay optimization in a kind of edge calculations environment | |
CN108494861A (en) | Method, apparatus and electronic equipment for Service Source allotment | |
CN109146381A (en) | Logistics data monitoring method, device, electronic equipment and computer storage medium | |
CN113408797A (en) | Method for generating flow-traffic prediction multi-time-sequence model, information sending method and device | |
CN106454437A (en) | Streaming media service rate prediction method and device | |
CN104580447A (en) | Spatio-temporal data service scheduling method based on access heat | |
CN109886445A (en) | A kind of tomorrow requirement prediction technique based on material requirements property quantification | |
CN110222996A (en) | Method for allocating tasks and device | |
CN108241534A (en) | A kind of task processing, distribution, management, the method calculated and device | |
KR20150116121A (en) | System and method for estimating value of continuous dependent variable, system and method for estimating billing of air cargo | |
CN107483355B (en) | Data center-oriented online scene low-bandwidth overhead traffic scheduling scheme | |
CN109784570B (en) | Intelligent workshop flexible production scheduling method based on information physical fusion system | |
CN104111875B (en) | Cloud data center increases number of tasks device for controlling dynamically, system and method newly | |
CN115113998A (en) | Digital twin resource allocation method and architecture | |
CN110717620A (en) | Order meal delivery time prediction method and device, electronic equipment and storage medium |
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 | ||
RJ01 | Rejection of invention patent application after publication |
Application publication date: 20190809 |
|
RJ01 | Rejection of invention patent application after publication |