CN107025223B

CN107025223B - A kind of buffer management method and server towards multi-tenant

Info

Publication number: CN107025223B
Application number: CN201610064482.6A
Authority: CN
Inventors: 文继军; 李健; 王伟
Original assignee: Huawei Technologies Co Ltd
Current assignee: XFusion Digital Technologies Co Ltd
Priority date: 2016-01-29
Filing date: 2016-01-29
Publication date: 2019-11-22
Anticipated expiration: 2036-01-29
Also published as: CN107025223A; WO2017128641A1

Abstract

The invention discloses a kind of buffer management method and server towards multi-tenant, this method comprises: according to the history buffer hit rate of target tenant in multi-tenant in default historical time section, according to the following buffer area hit rate of preset rules prediction target tenant；Compare the size of the following buffer area hit rate and the expectation buffer area hit rate of target tenant；If the following buffer area hit rate is greater than desired buffer area hit rate, then according to the difference of the following buffer area hit rate and the first pre-set buffer area hit rate, the first capacity is calculated according to the first preset algorithm, and discharges data page from the destination buffer of target tenant according to the first capacity；If the following buffer area hit rate is less than desired buffer area hit rate, according to the difference of the second pre-set buffer area hit rate and following buffer area hit rate, the second capacity is calculated according to the second preset algorithm, and be that data page is added in destination buffer according to the second capacity.The present invention is dynamically adapted the buffer pool size of target tenant.

Description

A kind of buffer management method and server towards multi-tenant

Technical field

The present invention relates to database technical field more particularly to a kind of buffer management methods and service towards multi-tenant Device.

Background technique

In the field database (Database, DB), tenant indicates " user " of a database, his " rental " database Part resource.Multi-tenant (Multi-Tenancy/Tenant) refers to that multiple tenants share a database instance.With cloud The rise of computing technique and flourish, multi-tenant application is deployed on cloud by more and more cloud service providers, multiple rents Family shares identical system or program assembly on cloud, shares resource (such as hardware resource, computing resource, memory source on cloud Deng), and can ensure that the isolation of data between tenant.For example, enterprise's mailbox is opened by certain site for service by multiple enterprises, each The authentication informations such as enterprise ID, password are filled in using same interface in Website login by enterprise, receive the same service flow in website Journey, but it is isolation that website, which is the information that each enterprise retains,.One enterprise can not see the information of other enterprises.Here, often A enterprise is exactly a tenant, includes multiple personal accounts of administrator and common employee in a tenant.

For multiple tenants when the data stored on to cloud access, there is bright for traditional buffer resource Managed Solution Aobvious deficiency: for each tenant when creating on cloud, cloud service system distributes the buffer area of fixed size on demand for it, due to system The buffer area of each tenant has uncertainty using size in the process of running, often because of the buffer pool size of initial setting up The wasting of resources that is excessive and causing buffer area, as the buffer pool size of initial setting up is too small and causes not to be able to satisfy tenant Demand.

Summary of the invention

The embodiment of the invention provides a kind of buffer management method and server towards multi-tenant.

In a first aspect, providing a kind of buffer management method towards multi-tenant, it is applied to but is not limited to server, example Such as database server, the data base handling system of large scale processing can be applied to, the data base handling system includes but not Be limited to: distributed disk base system, clustered database system and MPP (Massively Parallel Processing, Massive parallel processing) Database Systems.And the server can be common server, or Cloud Server (also known as cloud computing server or cloud host).The described method includes:

According to the history buffer hit rate of target tenant in default historical time section, the mesh is predicted according to preset rules The following buffer area hit rate of tenant is marked, the target tenant is one in the multi-tenant；

Compare the size of the following buffer area hit rate and the preset expectation buffer area hit rate of the target tenant, institute Stating desired buffer area hit rate is average demand value of the target tenant to buffer area hit rate, and the target tenant is servicing When applying for the registration of on device, expectation buffer area hit rate can be configured；

If the future buffer area hit rate is greater than expectation buffer area hit rate, ordered according to the following buffer area The difference of middle rate and target tenant preset first pre-set buffer area hit rate, calculates first according to the first preset algorithm Capacity, and data page is discharged from the corresponding destination buffer of the target tenant according to first capacity；Wherein described One pre-set buffer area hit rate is less than the following buffer area hit rate；

If the future buffer area hit rate is less than expectation buffer area hit rate, default according to the target tenant The second pre-set buffer area hit rate and the following buffer area hit rate difference, calculate second according to the second preset algorithm Capacity, and be that data page is added in the destination buffer according to second capacity；Wherein second pre-set buffer area is hit Rate is greater than the following buffer area hit rate.

By implement first aspect described in method, can according to the buffer area hit rate demand in target tenant's future with And the size relation of the preset buffer area hit rate demand of target tenant, dynamically adjust the corresponding target of the target tenant The capacity of buffer area.

In some implementations, described according to the following buffer area hit rate and the target tenant preset first The difference of pre-set buffer area hit rate calculates the first capacity according to the first preset algorithm, comprising:

Using the expectation buffer area hit rate as target tenant preset first pre-set buffer area hit rate, and root According to the difference and preset buffer pool size of the following buffer area hit rate and the expectation buffer area hit rate, according to the One preset algorithm calculates the first capacity.

It, can be the expectation buffer area hit rate demand for meeting the target tenant the case where by executing above-mentioned steps Under, it is calculated, is obtained using the difference of the following buffer area hit rate and the expectation buffer area hit rate of the target tenant The capacity for the data page having more than needed to the target tenant avoids so that the data page that the target tenant has more than needed be discharged The wasting of resources.

In some implementations, it is described according to target tenant preset second pre-set buffer area's hit rate with it is described The difference of the following buffer area hit rate calculates the second capacity according to the second preset algorithm, comprising:

Using the expectation buffer area hit rate as target tenant preset second pre-set buffer area hit rate, and root According to the difference and preset buffer pool size of the expectation buffer area hit rate and the following buffer area hit rate, according to the Two preset algorithms calculate the second capacity.

By executing above-mentioned steps, for the expectation buffer area hit rate demand for ensureing the target tenant, the mesh is utilized The expectation buffer area hit rate and the difference of the following buffer area hit rate for marking tenant calculate, and obtain the target tenant The capacity of data page required at present ensures the access of the target tenant to add data page for the target tenant It can demand.

Judge whether the following buffer area hit rate is greater than the preset maximum buffer hit rate of the target tenant, institute Maximum buffer hit rate is stated greater than expectation buffer area hit rate, the maximum buffer hit rate is the target tenant To the greatest requirements value of buffer area hit rate, the target tenant when applying for the registration of on the server, can to it is described most Big buffer pool size is configured；

If so, being hit the maximum buffer hit rate as the target tenant preset first pre-set buffer area Rate, and according to difference and the preset buffer area of the following buffer area hit rate and maximum preset buffer area hit rate Capacity calculates the first capacity according to the first preset algorithm.

It, can be in the premise for the maximum buffer hit rate demand for meeting the target tenant by executing above-mentioned steps Under, it is calculated using the following buffer area hit rate and the difference of the maximum buffer hit rate, obtains the target The capacity of buffer area data page more than needed avoids the wasting of resources so that the data page that the target tenant has more than needed be discharged.

In some implementations, it is described according to first capacity from the corresponding destination buffer of the target tenant Discharge data page, comprising:

Data page is discharged from the corresponding destination buffer of the target tenant according to first capacity, and will be released The data page of first capacity be added in freebuf, the data page in the freebuf is used for for described more The tenant for needing to add data page in tenant uses.

By executing above-mentioned steps, the data page released from the buffer area of the target tenant can be added to sky In not busy buffer area, the access performance demand of the target tenant can have not only been ensured, but also the target tenant can have been had more than needed slow It rushes area resource to release, to supply tenant's (including target that other in the multi-tenant need to increase buffer pool size The tenant that other tenants and the server in tenant, the multi-tenant newly increase) or server in other business It uses, improves the overall performance of multi-tenant and the resource utilization of buffer area.

Judge whether the following buffer area hit rate is greater than the preset maximum buffer hit rate of the target tenant, institute Maximum buffer hit rate is stated greater than expectation buffer area hit rate；

If it is not, then being hit the expectation buffer area hit rate as the target tenant preset first pre-set buffer area Rate, and held according to the difference and preset buffer area of the following buffer area hit rate and the expectation buffer area hit rate Amount, calculates the first capacity according to the first preset algorithm.

Data page is discharged from the corresponding destination buffer of the target tenant according to first capacity, it is described to release Data page be used to need to add for other in the multi-tenant tenant's use of data page, first capacity is less than or equal to Described other need to add tenant's data page capacity to be added of data page.

, can be when other tenants there is data page demand in the multi-tenant by executing above-mentioned steps, it will be from the mesh It marks the data page released in the buffer area of tenant and distributes to other tenants use, can both ensure the access of the target tenant Performance requirement, and the buffer resource that the target tenant has more than needed can be released, to supply other in the multi-tenant The tenant for needing to add data page uses, and improves the overall performance of multi-tenant and the resource utilization of buffer area.

Judge whether the following buffer area hit rate is less than the preset minimal buffering area hit rate of the target tenant, institute Minimal buffering area hit rate is stated less than expectation buffer area hit rate, minimal buffering area hit rate is the target tenant To the KB required of buffer area hit rate, the target tenant when applying for the registration of on the server, can to it is described most Minibuffer area hit rate is configured；

If so, being hit minimal buffering area hit rate as the target tenant preset second pre-set buffer area Rate, and held according to the difference and preset buffer area of minimal buffering area hit rate and the following buffer area hit rate Amount, calculates the second capacity according to the second preset algorithm.

By executing above-mentioned steps, for the minimal buffering area hit rate demand for ensureing the target tenant, the mesh is utilized The minimal buffering area hit rate and the difference of the following buffer area hit rate for marking tenant calculate, and obtain the target tenant It at least needs to add the data page of the second capacity, to add data page for the target tenant, ensures the target tenant's Access performance demand.

Judge whether the following buffer area hit rate is less than the preset minimal buffering area hit rate of the target tenant, institute Minimal buffering area hit rate is stated less than expectation buffer area hit rate；

If it is not, then being hit the expectation buffer area hit rate as the target tenant preset second pre-set buffer area Rate, and held according to the difference and preset buffer area of the expectation buffer area hit rate and the following buffer area hit rate Amount, calculates the second capacity according to the second preset algorithm.

In some implementations, the following buffer area hit rate and the preset expectation of target tenant The size of buffer area hit rate, comprising:

Judge that the following buffer area hit rate subtracts whether the difference for it is expected buffer area hit rate is greater than first in advance If threshold value, alternatively, judging that the expectation buffer area hit rate subtracts the difference of the following buffer area hit rate and whether is greater than the Two preset thresholds；

If it is default that the difference that the future buffer area hit rate subtracts the expectation buffer area hit rate is greater than described first Threshold value, then the following buffer area hit rate is greater than expectation buffer area hit rate；If expectation buffer area hit rate subtracts The difference of the following buffer area hit rate is gone to be greater than second preset threshold, then the following buffer area hit rate is less than institute State desired buffer area hit rate.It, can be with by the way that float value (i.e. described first preset threshold and second preset threshold) is arranged The capacity of destination buffer described in necessary adjustment is judged whether there is more flexiblely.

In some implementations, the basis presets the history buffer hit rate of target tenant in historical time section, The step for predicting the following buffer area hit rate of the target tenant according to preset rules is triggered by least one following event: Predetermined period reaches；Preset time point reaches；Number of the buffer area hit rate lower than the tenant of pre-set buffer area hit rate threshold value More than or equal to predetermined number threshold value.When meeting trigger event, server can understand the buffering of the target tenant in time Area's hit rate situation, and then the capacity of the destination buffer is adjusted in time.

Second aspect provides a kind of server, comprising:

Predicting unit, for the history buffer hit rate according to target tenant in default historical time section, according to default Rule predicts the following buffer area hit rate of the target tenant, and the target tenant is one in the multi-tenant；

Comparing unit, the following buffer area hit rate and the target predicted for the predicting unit The size of the preset expectation buffer area hit rate of tenant, the expectation buffer area hit rate are that the target tenant orders buffer area The average demand value of middle rate can be to the expectation buffer area hit rate when target tenant applies for the registration of on the server It is configured；

First administrative unit, if comparing the following buffer area hit rate for the comparing unit is greater than the expectation Buffer area hit rate is then hit according to the following buffer area hit rate and the target tenant preset first pre-set buffer area The difference of rate calculates the first capacity according to the first preset algorithm, and corresponding from the target tenant according to first capacity Destination buffer in discharge data page；Wherein first pre-set buffer area's hit rate is less than the following buffer area hit Rate；

Second administrative unit, if comparing the following buffer area hit rate for the comparing unit is less than the expectation Buffer area hit rate is then hit according to target tenant preset second pre-set buffer area's hit rate and the following buffer area The difference of rate calculates the second capacity according to the second preset algorithm, and is that the destination buffer adds according to second capacity Add data page；Wherein second pre-set buffer area's hit rate is greater than the following buffer area hit rate.

It, can buffer area hit rate demand according to target tenant's future and the target tenant by above-mentioned server The size relation of preset buffer area hit rate demand dynamically adjusts the capacity of the corresponding destination buffer of the target tenant.

In some implementations, first administrative unit is specifically used for:

By above-mentioned server, can in the case where meeting the expectation buffer area hit rate demand of the target tenant, It is calculated using the difference of the following buffer area hit rate and the expectation buffer area hit rate of the target tenant, obtains institute The capacity for the data page that target tenant has more than needed is stated, so that the data page that the target tenant has more than needed be discharged, avoids resource Waste.

In some implementations, second administrative unit is specifically used for:

The target is utilized by above-mentioned server for the expectation buffer area hit rate demand for ensureing the target tenant The expectation buffer area hit rate of tenant and the difference of the following buffer area hit rate calculate, and obtain the target tenant mesh The capacity of preceding required data page ensures the access performance of the target tenant to add data page for the target tenant Demand.

In some implementations, first administrative unit includes:

First judging unit, if comparing the following buffer area hit rate for the comparing unit is greater than the expectation Buffer area hit rate, then judge whether the following buffer area hit rate is greater than the preset maximum buffer life of the target tenant Middle rate, the maximum buffer hit rate are greater than expectation buffer area hit rate, and the maximum buffer hit rate is described For target tenant to the greatest requirements value of buffer area hit rate, the target tenant, can be with when applying for the registration of on the server The maximum buffer capacity is configured；

First computing unit, if it is described to judge that the following buffer area hit rate is greater than for first judging unit The preset maximum buffer hit rate of target tenant, then it is preset using the maximum buffer hit rate as the target tenant First pre-set buffer area hit rate, and according to the difference of following the buffer area hit rate and maximum preset buffer area hit rate Value and preset buffer pool size calculate the first capacity according to the first preset algorithm, and according to first capacity from institute It states in the corresponding destination buffer of target tenant and discharges data page.

By above-mentioned server, can under the premise of meeting the maximum buffer hit rate demand of the target tenant, It is calculated using the following buffer area hit rate and the difference of the maximum buffer hit rate, obtains the Target buffer The capacity of area's data page more than needed avoids the wasting of resources so that the data page that the target tenant has more than needed be discharged.

In some implementations, first administrative unit is specifically used for:

By above-mentioned server, the data page released from the buffer area of the target tenant can be added to the free time In buffer area, the access performance demand of the target tenant, but also the buffering that the target tenant can be had more than needed can have not only been ensured Area resource releases, with supply other in the multi-tenant need to increase buffer pool size tenant (including the target rent The tenant that other tenants and the server in family, the multi-tenant newly increase) or server in other business make With improving the overall performance of multi-tenant and the resource utilization of buffer area.

In some implementations, first administrative unit includes:

Second judgment unit, if comparing the following buffer area hit rate for the comparing unit is greater than the expectation Buffer area hit rate, then judge whether the following buffer area hit rate is greater than the preset maximum buffer life of the target tenant Middle rate, the maximum buffer hit rate are greater than expectation buffer area hit rate；

Second computing unit, if judging that the following buffer area hit rate is less than or waits for the second judgment unit In the preset maximum buffer hit rate of the target tenant, then using the expectation buffer area hit rate as the target tenant Preset first pre-set buffer area hit rate, and according to the following buffer area hit rate and the expectation buffer area hit rate Difference and preset buffer pool size calculate the first capacity according to the first preset algorithm, and according to first capacity from The target tenant discharges data page in corresponding destination buffer.

In some implementations, first administrative unit is specifically used for:

, can be when other tenants there is data page demand in the multi-tenant by above-mentioned server, it will be from the target The data page released in the buffer area of tenant distributes to other tenants use, can both ensure the access of the target tenant Energy demand, and the buffer resource that the target tenant has more than needed can be released, it is needed with supplying other in the multi-tenant The tenant for adding data page uses, and improves the overall performance of multi-tenant and the resource utilization of buffer area.

In some implementations, second administrative unit includes:

Third judging unit, if comparing the following buffer area hit rate for the comparing unit is less than the expectation Buffer area hit rate, then judge whether the following buffer area hit rate is less than the preset minimal buffering area life of the target tenant Middle rate, minimal buffering area hit rate are less than expectation buffer area hit rate, and minimal buffering area hit rate is described For target tenant to the KB required of buffer area hit rate, the target tenant, can be with when applying for the registration of on the server Minimal buffering area hit rate is configured；

Third computing unit, if it is described to judge that the following buffer area hit rate is less than for the third judging unit The preset minimal buffering area hit rate of target tenant, then it is preset using minimal buffering area hit rate as the target tenant Second pre-set buffer area hit rate, and according to the difference of minimal buffering area hit rate and the following buffer area hit rate with And preset buffer pool size, the second capacity is calculated according to the second preset algorithm, and be the mesh according to second capacity It marks buffer area and adds data page.

The target is utilized by above-mentioned server for the minimal buffering area hit rate demand for ensureing the target tenant The minimal buffering area hit rate of tenant and the difference of the following buffer area hit rate calculate, and obtain the target tenant extremely It needs to add the data page of the second capacity less, to add data page for the target tenant, ensures the visit of the target tenant Ask performance requirement.

In some implementations, second administrative unit includes:

4th judging unit, if comparing the following buffer area hit rate for the comparing unit is less than the expectation Buffer area hit rate, then judge whether the following buffer area hit rate is less than the preset minimal buffering area life of the target tenant Middle rate, minimal buffering area hit rate are less than expectation buffer area hit rate；

4th computing unit, if judging that the following buffer area hit rate is greater than or waits for the 4th judging unit In the preset minimal buffering area hit rate of the target tenant, then using the expectation buffer area hit rate as the target tenant Preset second pre-set buffer area hit rate, and according to the expectation buffer area hit rate and the following buffer area hit rate Difference and preset buffer pool size calculate the second capacity according to the second preset algorithm, and are according to second capacity Add data page in the destination buffer.

In some implementations, the comparing unit is specifically used for:

In some implementations, the predicting unit, which executes the basis and presets target tenant in historical time section, goes through History buffer area hit rate, according to preset rules predict following buffer area hit rate this operation of the target tenant by with down toward Few event triggering: predetermined period reaches；Preset time point reaches；Buffer area hit rate is lower than pre-set buffer area hit rate threshold The number of the tenant of value is greater than or equal to predetermined number threshold value.When meeting trigger event, server can understand described in time The buffer area hit rate situation of target tenant, and then the capacity of the destination buffer is adjusted in time.

The third aspect provides a kind of server, including processor and memory, wherein the server can be counted for example According to library server, the data base handling system of large scale processing can be applied to, the data base handling system includes but is not limited to: Distributed disk base system, clustered database system and MPP Database Systems.And the server can be generic services Device, or Cloud Server；The memory is for storing the buffer management code towards multi-tenant, the processing Device is used to that the program code of the memory storage to be called to execute above-mentioned first aspect or any implementation institute of first aspect The buffer management method towards multi-tenant of description.

In some implementations of the invention, first preset algorithm are as follows: C1=(A-B) * D, wherein C1 is described First capacity, A are the following buffer area hit rate, and B is first pre-set buffer area hit rate, and D is described preset slow Rush Qu Rongliang.The data page capacity that the destination buffer can release can be calculated by the first preset algorithm.

In some implementations of the invention, first preset algorithm are as follows: C1=min ((A-B) * D, E), wherein C1 is first capacity, and A is the following buffer area hit rate, and B is first pre-set buffer area hit rate, and D is described Preset buffer pool size, E are the data page capacity that other tenants are to be added in the multi-tenant.By by the target tenant The data page capacity that the can release data page capacity to be added with other tenants is compared, it may be determined that goes out the Target buffer The practical data page capacity discharged in area.

In some implementations of the invention, second preset algorithm are as follows: C2=min ((F-A) * D, G), wherein C2 is second capacity, and F is second pre-set buffer area hit rate, and A is the following buffer area hit rate, and D is described Preset buffer pool size, G are the number in other tenants are to be released in the multi-tenant data page capacity and freebuf According to the sum of page capacity.Pass through server currently available data page capacity and target tenant data page capacity to be added It is compared, it may be determined that go out the data page capacity of the destination buffer actual interpolation.

In some implementations of the invention, the preset buffer pool size includes: that the target tenant is preset Minimal buffering area capacity, the preset expectation buffer pool size of the target tenant, the preset maximum buffer of target tenant Current at least one of the capacity of capacity and the destination buffer.When the target tenant applies for the registration of on the server, Minimal buffering area capacity, the expectation buffer pool size and the maximum buffer capacity can be configured respectively.

In some implementations of the invention, the default historical time section can be a past period, It can be past at least two period.For example, when the default historical time section can be by the server current prediction Between put and the time interval shortest historical forecast time point with the current predictive time point composed by the period, It can be every two adjacent predicted time point between the current predicted time point and preset predicted time point to be formed P (P is positive integer, and P >=1) a period, can also (Q be positive integer, and 1≤Q for any Q in the P period ≤ P) a period.Wherein, the preset predicted time point may include but be not limited to server starting postrun the One predicted time point.

In some implementations of the invention, the preset rules include: that (N is positive whole according to the N of the target tenant Number, N >=1) a history buffer hit rate, the average buffer hit rate of N number of history buffer hit rate is calculated, by institute State the following buffer area hit rate that average buffer hit rate is determined as the target tenant；Alternatively, according to the target tenant M (M is positive integer, M >=2) a history buffer hit rate variation tendency, predict the following buffer area of the target tenant Hit rate.

Implement the embodiment of the present invention, according to the following buffer area hit rate of the target tenant in the obtained multi-tenant of prediction with And the preset buffer area hit rate of target tenant, dynamically to adjust the corresponding destination buffer of target tenant in multi-tenant Capacity, not only can be to avoid the access performance for influencing tenant because of out of buffers, but also can be made to avoid because buffer area is excessive At the wasting of resources.

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 described.

Fig. 1 is the configuration diagram of data base buffer in server provided in an embodiment of the present invention；

Fig. 2 is a kind of flow diagram of buffer management method towards multi-tenant provided in an embodiment of the present invention；

Fig. 3 is a kind of structural schematic diagram of server provided in an embodiment of the present invention；

Fig. 4 is the structural schematic diagram of another server provided in an embodiment of the present invention；

Fig. 5 is the structural schematic diagram of another server provided in an embodiment of the present invention；

Fig. 6 is the structural schematic diagram of another server provided in an embodiment of the present invention；

Fig. 7 is the structural schematic diagram of another server provided in an embodiment of the present invention；

Fig. 8 is the structural schematic diagram of another server provided in an embodiment of the present invention.

Specific embodiment

Below in conjunction with attached drawing, the embodiment of the present invention is described.

It should be noted that the term used in embodiments of the present invention is only merely for the mesh of description specific embodiment , it is not intended to limit the invention." the one of the embodiment of the present invention and singular used in the attached claims Kind ", " described " and "the" are also intended to including most forms, unless the context clearly indicates other meaning.It is also understood that this Term "and/or" used herein refers to and includes one or more associated any or all possible group for listing project It closes.Although in embodiments of the present invention various functional modules may be described using term " first ", " second " and " third " etc. Or unit, but these proper nouns, functional module or unit should not necessarily be limited by these terms.These terms are only used to each function mould Block or unit are distinguished from each other out, and there is no the meanings that tandem limits.In addition, " multi-tenant " in the embodiment of the present invention includes At least two tenants.

Server described in following embodiment is mountable database management system software, is specifically as follows data Library server can be applied to the data base handling system of large scale processing, and the data base handling system includes but is not limited to: point Cloth disk base system, clustered database system and MPP Database Systems.And the server can be generic services Device, or Cloud Server.Wherein, common server refers to independent server, is specifically as follows one or more calculating Machine can be disposed and provide service in a local network for multiple users (can be regarded as tenant), be stored in the common server multiple The respective user data of user, multiple users can access the common server by local area network；Cloud Server is based on cloud framework Server cluster, have a multiple servers, beyond the clouds, multiple tenants rent computing resource, storage resource etc. from cloud for deployment.It should Be stored with the respective tenant data of multiple tenants in Cloud Server, multiple tenants can by the internet access Cloud Server, In, these services include but is not limited to: inquiry, update, storage etc..

First introduce the general plotting of buffer management scheme provided by the invention below, the present invention will be described in detail side subsequent again The detailed step of case.Firstly, being that the framework of data base buffer in server provided in an embodiment of the present invention shows referring to Figure 1 It is intended to, is illustrated with reference to Fig. 1 the general plotting of the buffer management method provided by the invention towards multi-tenant: in Fig. 1, service Device is towards 4 tenants, and the mark of 4 tenants is respectively T1, T2, T3 and T4, and 4 tenants infused on that server by application Volume, and 4 tenants are registered in the server when creation, server is respectively 4 tenants point from the memory of operating system The privately owned buffer area of certain capacity is matched, here for the ease of distinguishing, the title of 4 privately owned buffer areas has been respectively designated as The capacity of Buffer1, Buffer2, Buffer3 and Buffer4,4 privately owned buffer areas may be the same or different, it is assumed that 4 The capacity of a buffer area is respectively 2GB (Gigabyte, gigabytes), 2GB, 1GB and 3GB.In addition, server is also in the present invention Certain shared resource is set, i.e., freebuf shown in FIG. 1.

Wherein, freebuf is made of several idle data pages, which can be the number of server reserves According to page, it is also possible to the empty data page released from the buffer area of certain tenants.The present invention is combined using dynamic and static state Mode use buffer area, when initial server be each tenant distribute certain capacity buffer resource, server run In the process, server can be according to the pre-set performance requirement of each tenant (specially each pre-set buffer area of tenant Hit rate) and each tenant actual performance (specially currently for each tenant prediction the following buffer area hit rate), The buffer pool size of dynamic adjustment each tenant (including reducing and increasing), is suitably released from the buffer area of the higher tenant of performance Partial buffer area is added into freebuf, and from freebuf for the lower tenant of performance in discharge section buffer area, To improve the resource utilization of buffer area while the overall performance for meeting multi-tenant.

It should be noted that buffer area is made of several data pages (alternatively referred to as data block), a data page Capacity be usually 4KB (Kilobyte, kilobytes) or 8KB, continuous data page constitutes a buffer area, the appearance of buffer area Amount refers to the sum of the capacity of each data page in the buffer area.

It should be noted that data page in data page and freebuf in the buffer area of 4 tenants shown in FIG. 1 Pattern difference is intended merely to as differentiation, and the data page in the buffer area of 4 tenants is stored with the corresponding data of each tenant, should Class data page is used to keep in the data that server is read in from the hardware stores node such as disk or the hardware such as disk will be written and deposits Store up the data of node.And the data page in freebuf is idle data page, it is generally the case that will not in such data page Be stored with the data of tenant, avoid such data page by other tenants using when inquire upper one data using tenant, from And enhance the Information Security of each tenant.

Following embodiment is the explanation carried out by taking the target tenant in multi-tenant as an example, for its in the multi-tenant His tenant can also equally realize that dynamic is adjusted using the buffer management method provided by the embodiment of the present invention towards multi-tenant The buffer pool size of other whole tenants.

Below with reference to Fig. 2 specific embodiment that the present invention will be described in detail.

Fig. 2 is referred to, is that a kind of process of buffer management method towards multi-tenant provided in an embodiment of the present invention is shown It is intended to.The buffer management method towards multi-tenant can be applied to but be not limited to server, such as database server, and institute Stating server can be common server, or Cloud Server, should buffer management method towards multi-tenant include but It is not limited to following steps.

S201: it according to the history buffer hit rate of target tenant in default historical time section, is predicted according to preset rules The following buffer area hit rate of the target tenant.

Specifically, the target tenant is one in multi-tenant.In server operational process, the server is by institute The history buffer hit rate for stating target tenant is hit according to the following buffer area that preset rules predict to obtain the target tenant Rate.Wherein, the history buffer hit rate refers to reads from data base buffer in default historical time section server The ratio of the total quantity of data is read to the quantity of data and the server, the total quantity that the server reads data includes From the data bulk read in the data page of data base buffer and the number read from the hardware stores node such as disk The sum of data bulk.

Optionally, the history buffer hit rate of the target tenant is predicted to obtain by the server according to preset rules The step for following buffer area hit rate of the target tenant can be triggered by least one following event: predetermined period reaches (i.e. cyclic forecast)；Preset time point reaches (i.e. timing predictions)；Buffer area hit rate is lower than pre-set buffer area hit rate threshold The number of the tenant of value is greater than or equal to predetermined number threshold value.Wherein, the predetermined period, the preset time point, described Pre-set buffer area hit rate threshold value and the predetermined number threshold value can be arranged by the system default of the server, can also be by institute Administrator's manual setting of server is stated, the embodiment of the present invention is not especially limited.When meeting trigger event, server can be with The following buffer area hit rate situation of the target tenant is understood in time, and then the buffer area of the target tenant is held in time Amount is adjusted.

For example, the predetermined period is 30 minutes, then the server is primary every prediction in 30 minutes in the process of running The following buffer area hit rate of the target tenant；Alternatively, the preset time point can be each integral point (such as 3 in one day Point, 7 points, 8 points etc.) and each least bit (such as 3 points 30 minutes, 7 points 30 minutes, 8 points 30 minutes etc.), then the server is in each integral point Or the following buffer area hit rate of the primary target tenant is predicted when least bit arrival；Alternatively, pre-set buffer area hit rate Threshold value is 80%, and the predetermined number threshold value is 5, then the server is in the rent for detecting that buffer area hit rate is lower than 80% When the number at family is greater than or equal to 5, the following buffer area hit rate of the target tenant is predicted.

Optionally, the default historical time section can be a past period, or past at least two A period, the embodiment of the present invention are not especially limited.For example, the default historical time section can be current by the server When composed by predicted time point and time interval shortest historical forecast time point with the current predictive time point Between section, be also possible to every two adjacent predicted time point between the current predicted time point and preset predicted time point Composed P (P is positive integer, and P >=1) a period, can also in the P period any Q (Q is positive integer, And 1≤Q≤P) a period.Wherein, the preset predicted time point may include but be not limited to the server starting operation First predicted time point afterwards.For example, the server currently carries out the following buffer area hit rate of the target tenant Time point when prediction is 11 points of the morning, is the morning 10 with 11 points of time interval of a morning shortest historical forecast time point Point 30 minutes, then the default historical time section is 10:30:00-11:00:00；Alternatively, the predicted time that the server is current Point is 11 points of the morning, and preset predicted time point is 8 a.m., and predetermined period of the server setting is half an hour, then institute State default historical time section be respectively 08:00:00-08:30:00,08:30:00-09:00:00,09:00:00-09:30:00, 09:30:00-10:00:00,10:00:00-10:30:00 and 10:30:00-11:00:00；Alternatively, the default historical time Section is any three in above-mentioned 6 periods: 08:00:00-08:30:00,09:00:00-09:30:00 and 10:00:00- 10:30:00。

Optionally, the preset rules include but is not limited to: according to the N of the target tenant (N is positive integer, 1≤N≤ P) a history buffer hit rate calculates the average buffer hit rate of N number of history buffer hit rate, will be described average Buffer area hit rate is determined as the following buffer area hit rate of the target tenant；Alternatively, according to the M of the target tenant, (M is Positive integer, 2≤M≤P) a history buffer hit rate variation tendency, predict the following buffer area hit of the target tenant Rate.

For example, the default historical time section is 10:30:00-11:00:00, the server is counted at 11 points of the morning History buffer hit rate to the target tenant in this period of 10:30:00-11:00:00 is 95%, then by 95% It is determined as the following buffer area hit rate of the target tenant；Alternatively, the default historical time section is respectively 08:00:00- 08:30:00、08:30:00-09:00:00、09:00:00-09:30:00、09:30:00-10:00:00、10:00:00-10: 30:00 and 10:30:00-11:00:00, the server obtain the target tenant at 11 points of the morning in above-mentioned 6 history Between history buffer hit rate in section be respectively 95%, 90%, 85%, 95%, 85%, 90%, then calculate 6 history bufferings The average buffer hit rate 90% being calculated is determined as the target tenant by the average buffer hit rate of area's hit rate The following buffer area hit rate；Alternatively, the default historical time section is respectively 08:00:00-08:30:00,08:30:00- 09:00:00,09:00:00-09:30:00,09:30:00-10:00:00,10:00:00-10:30:00 and 10:30:00-11: 00:00, the server obtain history buffer life of the target tenant in above-mentioned 6 historical time sections at 11 points of the morning Middle rate is respectively 95%, 90%, 85%, 95%, 85%, 90%, then according to the variation tendency of 6 history buffer hit rates, The following buffer area hit rate for being determined as the target tenant for 95%.

S202: the future buffer area hit rate is big with the preset expectation buffer area hit rate of the target tenant It is small.

Specifically, the expectation buffer area hit rate is average demand value of the target tenant to buffer area hit rate, The target tenant can be configured the expectation buffer area hit rate when applying for the registration of in the server, the mesh Expectation buffer area hit rate can be only arranged in mark tenant, three buffer area hit rates can also be arranged simultaneously, respectively most Big buffer area hit rate, expectation buffer area hit rate and minimal buffering area hit rate, wherein the maximum buffer hit rate > Minimal buffering area hit rate described in the expectation buffer area hit rate >, the maximum buffer hit rate and the minimal buffering Area's hit rate can be described in detail subsequent.The server is according to the following buffer area hit rate of the target tenant and institute The size relation between desired buffer area hit rate is stated to determine whether needing to adjust the corresponding Target buffer of the target tenant The capacity in area.If the future buffer area hit rate is greater than the expectation buffer area hit rate or the following buffer area life Middle rate is less than expectation buffer area hit rate, then the result judged adjusts the capacity of the destination buffer as needs It is whole, it, can be to the appearance of the destination buffer if the future buffer area hit rate is equal to expectation buffer area hit rate Whether amount is without adjustment, or the capacity to the destination buffer is needed to be adjusted determine according to actual needs.

Optionally, the following buffer area hit rate and the preset expectation buffer area of the target tenant are hit The size of rate, comprising:

Judge that the following buffer area hit rate subtracts whether the difference for it is expected buffer area hit rate is greater than first in advance If threshold value, alternatively, judging that the expectation buffer area hit rate subtracts the difference of the following buffer area hit rate and whether is greater than the Two preset thresholds；If it is pre- that the difference that the future buffer area hit rate subtracts the expectation buffer area hit rate is greater than described first If threshold value, then the following buffer area hit rate is greater than expectation buffer area hit rate；If expectation buffer area hit rate The difference for subtracting the following buffer area hit rate is greater than second preset threshold, then the following buffer area hit rate is less than Expectation buffer area hit rate.

Wherein, first preset threshold and second preset threshold can be defaulted by the operating system of the server sets It sets, can also be by administrator's manual setting of the server, and first preset threshold and second preset threshold can With identical, can also be different, the embodiment of the present invention is not especially limited.By the way that float value (i.e. described first preset threshold is arranged With second preset threshold), the capacity of destination buffer described in necessary adjustment can be judged whether there is more flexiblely.Example Such as, first preset threshold is 2%, and the future buffer area hit rate is 95%, and expectation buffer area hit rate is 90%, the difference that the future buffer area hit rate subtracts the expectation buffer area hit rate is greater than 2%, then the result judged is The future buffer area hit rate is greater than expectation buffer area hit rate；Alternatively, second preset threshold is 2%, it is described The following buffer area hit rate is 85%, and the expectation buffer area hit rate is 90%, and expectation buffer area hit rate subtracts institute The difference for stating the following buffer area hit rate is greater than 2%, then the result judged is less than the phase as the following buffer area hit rate Hope buffer area hit rate.

S203: if the future buffer area hit rate is greater than expectation buffer area hit rate, according to described following slow The difference for rushing area's hit rate Yu target tenant preset first pre-set buffer area hit rate is calculated according to the first preset algorithm First capacity out, and data page is discharged from the corresponding destination buffer of the target tenant according to first capacity；Wherein First pre-set buffer area's hit rate is less than the following buffer area hit rate.

Specifically, if the following buffer area hit rate of the target tenant is greater than expectation buffer area hit rate, benefit With the following buffer area hit rate of the target tenant and the difference of target tenant preset first pre-set buffer area hit rate Value is calculated, and the capacity for the data page that the target tenant has more than needed is obtained, thus according to first capacity by the target The data page that tenant has more than needed discharges, and avoids the wasting of resources.Wherein, the data page released from the destination buffer can Need to increase the tenant of buffer pool size (including its in the target tenant, the multi-tenant for other in the multi-tenant The tenant that his tenant and the server newly increase) or server in other business use.

Optionally, first preset algorithm are as follows: C1=(A-B) * D, wherein C1 is first capacity, A be it is described not Carry out buffer area hit rate, B is first pre-set buffer area hit rate, and D is the preset buffer pool size.Wherein, described Preset buffer pool size includes: the preset minimal buffering area capacity of the target tenant, the preset expectation of target tenant In buffer pool size, the current capacity of the preset maximum buffer capacity of the target tenant and the destination buffer at least It is a kind of.It, can be slow to minimal buffering area capacity, the expectation respectively when the target tenant applies for the registration of on the server It rushes area's capacity and the maximum buffer capacity is configured, can also be set just for one of buffer pool size It sets.For example, the minimal buffering area capacity of the target tenant setting is 1GB, it is expected that buffer pool size is 1.5GB, maximum cushioning Area's capacity is 2GB.

Optionally, first preset algorithm are as follows: C1=min ((A-B) * D, E), wherein C1 is first capacity, A For the following buffer area hit rate, B is first pre-set buffer area hit rate, and D is the preset buffer pool size, root The available data page capacity of the target tenant can be calculated according to formula (A-B) * D, E is other tenants in the multi-tenant Data page capacity to be added.It waits adding if the available data page capacity of target tenant is greater than or equal to other described tenants The data page capacity added, then first capacity is equal to other described tenants data page capacity to be added；If the target is rented The available data page capacity in family is less than other described tenants data page capacity to be added, then first capacity is equal to described The available data page capacity of target tenant.So as to determine that the destination buffer is practical according to the demand of other tenants The data page capacity for needing to discharge can either ensure the performance requirement of the target tenant, and data page can be supplied to other There is the tenant of demand to use, improves the overall performance of multi-tenant and the utilization rate of buffer area.

Optionally, described according to the following buffer area hit rate and target tenant preset first pre-set buffer area The difference of hit rate calculates the first capacity according to the first preset algorithm, comprising:

Specifically, if the target tenant is only arranged desired buffer area hit rate in initial setting up, if Judge that the following buffer area hit rate of the target tenant is greater than expectation buffer area hit rate, then may indicate that in future A period of time server can satisfy the expectation buffer area hit rate demand of the target tenant in this case can To be calculated using the following buffer area hit rate of the target tenant and the difference of the expectation buffer area hit rate, obtain The capacity for the data page that the target tenant has more than needed is the first capacity, thus the number for the first capacity that the target tenant is had more than needed It is discharged according to page, avoids the wasting of resources.

For example, the future buffer area hit rate is 95%, the expectation buffer area hit rate is 90%, and the target is slow It rushes the current capacity in area and calculates the first capacity C 1=then according to the first preset algorithm C1=(A-B) * D for 2GB (95%-90%) * 2GB=102MB (Megabyte, Mbytes).

Specifically, the maximum buffer hit rate is greatest requirements value of the target tenant to buffer area hit rate, The target tenant can be configured the maximum buffer hit rate when applying for the registration of on the server, if The target tenant is respectively arranged desired buffer area hit rate and maximum buffer hit rate in initial setting up, if Judge that the following buffer area hit rate of the target tenant is greater than expectation buffer area hit rate, then needs further to judge Whether the future buffer area hit rate is greater than the maximum buffer hit rate, if so, may indicate that at following one section Time server can satisfy the maximum buffer hit rate demand of the target tenant, in such a case, it is possible to utilize The difference of the following buffer area hit rate and the maximum buffer hit rate of the target tenant calculates, and obtains the mesh The capacity for the data page that mark tenant has more than needed is the first capacity, and first capacity is that the target tenant currently can at least discharge The data page that the target tenant has more than needed is discharged, avoids the wasting of resources by the capacity of data page out.

For example, the future buffer area hit rate is 95%, the expectation buffer area hit rate is 90%, described maximum slow Rushing area's hit rate is 94%, and the preset minimal buffering area capacity of target tenant is 1GB, then according to first preset algorithm C1=(A-B) * D calculates the first capacity C 1=(95%-94%) * 1GB=10*0MB.

Optionally, if judging, the following buffer area hit rate of the target tenant is hit greater than the maximum buffer Rate, then the maximum buffer hit rate that may indicate that server can satisfy the target tenant in the coming period of time need It asks, in the case where meeting the expectation buffer area hit rate demand of the target tenant, also can use the target tenant's The difference of the following buffer area hit rate and the expectation buffer area hit rate is calculated, and show that the target tenant can release Data page be the first capacity, first capacity indicates the appearance for the data page that the target tenant currently can at most release Amount, the data page that the target tenant has more than needed is discharged, the wasting of resources is avoided.

For example, the future buffer area hit rate is 95%, the expectation buffer area hit rate is 90%, described maximum slow Rushing area's hit rate is 94%, and the preset minimal buffering area capacity of target tenant is 1GB, then according to first preset algorithm C1=(A-B) * D calculates the first capacity C 1=(95%-90%) * 1GB=51.2MB.

Optionally, described to discharge data from the corresponding destination buffer of the target tenant according to first capacity Page, comprising:

Specifically, discharging the target according to first capacity being calculated after calculating first capacity The data page of buffer area, and the data page released is added in the freebuf of the server.The free buffer Data page in area is for needing to add the tenant of data page or other business use of the server, to improve rent more The overall performance at family and the utilization rate of buffer area, wherein the tenant for needing to add data page includes but is not limited to: the mesh The tenant that mark tenant, other tenants in the multi-tenant and the server newly increase.

For example, if first capacity is 10.2MB, if the capacity of each data page is 8KB, the Target buffer Area needs to discharge 1306 data pages of 10.2MB/8KB ≈, and 1306 data pages discharged from the destination buffer are added Into freebuf.

Optionally, if only desired buffer area hit rate be arranged when the target tenant is initial, work as judgement When the following buffer area hit rate is greater than the expectation buffer area hit rate out, it can will be discharged from the destination buffer The data page of the first capacity out is added in freebuf；Alternatively, if being buffered when the target tenant is initial to expectation Area's hit rate and maximum buffer hit rate are arranged respectively, then when judging that the following buffer area hit rate is greater than institute When stating maximum buffer hit rate, the data page of the first capacity released from the destination buffer can be added to sky In not busy buffer area, so that the data page that the target tenant has more than needed be released, the wasting of resources is avoided.

Specifically, if the target tenant is in initial setting up respectively to desired buffer area hit rate and maximum buffer Hit rate is arranged, if judging, the following buffer area hit rate of the target tenant is hit greater than the expectation buffer area Rate, then need further to judge whether the following buffer area hit rate is greater than the maximum buffer hit rate, if it is not, then may be used To show that server in the coming period of time is not able to satisfy the maximum buffer hit rate demand of the target tenant, still It can satisfy the expectation buffer area hit rate demand of the target tenant, in such a case, it is possible to utilize the target tenant The following buffer area hit rate and it is described expectation buffer area hit rate difference calculated, obtain what the target tenant had more than needed The capacity of data page is the first capacity, and the data page that the target tenant has more than needed is discharged, the wasting of resources is avoided.

For example, the future buffer area hit rate is 95%, the expectation buffer area hit rate is 90%, described maximum slow Rushing area's hit rate is 96%, and the preset expectation buffer pool size of target tenant is 1.5GB, then according to the described first pre- imputation Method C1=(A-B) * D calculates the first capacity C 1=(95%-90%) * 1.5GB=76.8MB.

Specifically, discharging the target according to first capacity being calculated after calculating first capacity The data page of buffer area, and the data page released is added to the corresponding buffer area of tenant that other need to add data page In, to supply other tenants use.To provide partial data page under the premise of ensureing the access performance of the target tenant It needs the tenant of data page to use to other, improves the utilization rate of buffer area.

It should be noted that remaining in data page and the freebuf when judging that other tenants need to add When data page deficiency, first capacity is calculated, and discharge the data page in the destination buffer according to first capacity, In the buffer area of other tenants described in being added to from the data page released in the destination buffer, so as to according to it The demand of his tenant discharges the data page in the destination buffer, between different tenants dynamic adaptation data page, favorably In the utilization rate for improving buffer area.

For example, the future buffer area hit rate is 95%, the expectation buffer area hit rate is 90%, described maximum slow Rushing area's hit rate is 96%, and the preset it is expected buffer pool size of target tenant is 1.5GB, and there are two rents in the multi-tenant Family needs to add data page: the first tenant needs to add the data page of 10MB, and the second tenant needs to add the data page of 20MB, and There is no remaining data page in the freebuf at this time, then according to the first preset algorithm C1=min ((A-B) * D, E), calculating the data page capacity that the target tenant can provide out is (A-B) * D=(95%-90%) * 1.5GB= 76.8MB, D=10MB+20MB=30MB then calculate the first capacity C 1=(76.8MB, the 30MB)=30MB.Then from institute The data page for discharging 30MB in destination buffer is stated, and 10MB data page therein is added to the corresponding buffer area of the first tenant In, the data page of other 20MB is added in the corresponding buffer area of the second tenant.

It should be noted that the step for data page of the release destination buffer described in the embodiment of the present invention Specifically: first the data in data page to be released are purged, then be added in the freebuf (or add It is added in the corresponding buffer area of tenant that other need to add data page, or other regions being added in server memory), And before removing the data in a certain data page, it can first judge whether the data in the data page have been updated and (repair Change) or the data page in whether increase new data, if so, needing that the data in the data page are first write disk etc. hard In part memory node, the data in hardware store node are updated with realizing, it is then again that the data in the data page are clear It removes, the data of the target tenant is accessed when avoiding other tenants subsequent using the data page, to improve the number of multi-tenant According to safety.

S204: it if the future buffer area hit rate is less than expectation buffer area hit rate, is rented according to the target The difference of family preset second pre-set buffer area's hit rate and the following buffer area hit rate, calculates according to the second preset algorithm Second capacity out, and be that data page is added in the destination buffer according to second capacity；Wherein second pre-set buffer Area's hit rate is greater than the following buffer area hit rate.

Specifically, if the following buffer area hit rate of the target tenant is less than expectation buffer area hit rate, benefit It is calculated with the difference of the expectation buffer area hit rate and the following buffer area hit rate, obtains the target tenant mesh The capacity of preceding required data page ensures the access performance of the target tenant to add data page for the target tenant Demand.

It is understood that the total capacity of data page can generate vital shadow to access performance in destination buffer It rings, the total capacity of data page is bigger in destination buffer, and cacheable data are more, and target tenant passes through server access data When, the probability that the server inquires data from buffer area is bigger, and response speed is faster, and access performance is also better.

Specifically, second preset algorithm are as follows: C2=min ((F-A) * D, G), wherein C2 is second capacity, F For second pre-set buffer area hit rate, A is the following buffer area hit rate, and D is the preset buffer pool size, (F-A) * D is target tenant data page capacity to be added, and G is the data page that other tenants are to be released in the multi-tenant The sum of data page capacity in capacity and freebuf.If target tenant data page capacity to be added is greater than or equal to The sum of data page capacity in the multi-tenant in other tenants data page capacity and freebuf to be released, then described Two capacity be equal to the data page capacity in the multi-tenant in other tenants data page capacity and freebuf to be released it With；If target tenant data page capacity to be added is less than the data page capacity that other tenants are to be released in the multi-tenant With the sum of the data page capacity in freebuf, then second capacity is equal to target tenant data page to be added and holds Amount.

Wherein, the preset buffer pool size includes: the preset minimal buffering area capacity of the target tenant, the mesh The preset expectation buffer pool size of mark tenant, the preset maximum buffer capacity of the target tenant and the destination buffer are worked as At least one of preceding capacity.It, can be respectively to the minimal buffering when target tenant applies for the registration of on the server Area's capacity, the expectation buffer pool size and the maximum buffer capacity are configured.

Optionally, described according to target tenant preset second pre-set buffer area's hit rate and the following buffer area The difference of hit rate calculates the second capacity according to the second preset algorithm, comprising:

Specifically, if the target tenant is only arranged desired buffer area hit rate in initial setting up, if Judge that the following buffer area hit rate of the target tenant is less than expectation buffer area hit rate, then may indicate that in future A period of time server be not able to satisfy the expectation buffer area hit rate demand of the target tenant, to ensure that the target is rented The expectation buffer area hit rate demand at family can use the expectation buffer area hit rate and the following buffering of the target tenant The difference of area's hit rate is calculated, and obtains the target tenant and need the capacity of data page to be added to be the second capacity, thus The data page that the second capacity is added for the target tenant, ensures the access performance demand of the target tenant.

For example, the future buffer area hit rate is 85%, the expectation buffer area hit rate is 90%, and the target is slow Rushing the current capacity in area is 2GB, other tenants data page total capacity to be released and the freebuf in the multi-tenant In the sum of data page capacity be 20MB+1GB=1044MB, then according to the second preset algorithm C2=mnn ((F-A) * D, G), target tenant data page capacity to be added is calculated are as follows: (F-A) * D=(90%-85%) * 2GB=102MB, then The second capacity C 2=min (102MB, the 1044MB)=102MB is calculated, then adds 102MB's for the destination buffer Data page, since the data page in the freebuf is supplied to the target tenant enough, then by the freebuf The data page of middle 102MB is added in the destination buffer.

Specifically, minimal buffering area hit rate is KB required of the target tenant to buffer area hit rate, The target tenant can be configured minimal buffering area hit rate when applying for the registration of on the server, if The target tenant is arranged desired buffer area hit rate and minimal buffering area hit rate in initial setting up, if sentencing The following buffer area hit rate of the disconnected target tenant out is less than expectation buffer area hit rate, then needs further to judge institute State whether the following buffer area hit rate is less than minimal buffering area hit rate, if so, may indicate that at following one section Between server be not able to satisfy the minimal buffering area hit rate demand of the target tenant, for the minimum for ensureing the target tenant Buffer area hit rate demand can use the minimal buffering area hit rate and the following buffer area hit rate of the target tenant Difference calculated, obtain the data page that the target tenant at least needs to add the second capacity, thus for the target rent Data page is added at family, ensures the access performance demand of the target tenant.

For example, the future buffer area hit rate is 80%, the expectation buffer area hit rate is 90%, described minimum slow Rushing area's hit rate is 85%, and the preset minimal buffering area capacity is 1GB, other tenants number to be released in the multi-tenant It is 50MB+20MB=70MB according to the sum of the data page capacity in page total capacity and the freebuf, then according to described second Preset algorithm C2=min ((F-A) * D, G), calculating target tenant data page capacity to be added is (85%-80%) * 1GB=51.2MB then calculates the second capacity C 2=min (51.2MB, the 70MB)=51.2MB.It then can preferentially will be described Remaining 20MB data page is added in the destination buffer in freebuf, and from the tenant of other releasable data pages The data page of 31.2MB is discharged in corresponding buffer area, and the data page of the 31.2MB released is added to the Target buffer Qu Zhong.

Specifically, if the target tenant hits desired buffer area hit rate and minimal buffering area in initial setting up Rate is arranged, if judging, the following buffer area hit rate of the target tenant is hit less than the expectation buffer area Rate, then need further to judge whether the following buffer area hit rate is less than minimal buffering area hit rate, if it is not, then may be used To show that server in the coming period of time can satisfy the minimal buffering area hit rate of the target tenant, but cannot Meet the expectation buffer area hit rate demand of the target tenant, to ensure that the expectation buffer area hit rate of the target tenant needs It asks, the expectation buffer area hit rate and the difference of the following buffer area hit rate that can use the target tenant are counted It calculates, obtains the data page that the target tenant at least needs to add the second capacity, so that data page is added for the target tenant, Ensure the access performance demand of the target tenant.

For example, the future buffer area hit rate is 85%, the expectation buffer area hit rate is 90%, described minimum slow Rushing area's hit rate is 80%, and the preset expectation buffer pool size of target tenant is 1.5GB, other tenants in the multi-tenant The sum of data page capacity in data page total capacity and the freebuf to be released is 50MB+0MB=50MB, then basis It is (F- that the second preset algorithm C2=min ((F-A) * D, G), which calculates target tenant data page capacity to be added, A) * D=(90%-85%) * 1.5GB=76.8MB then calculates the second capacity C 2=min (76.8MB, 50MB)=50MB.By There is no remaining data page in the freebuf, is then released from the corresponding buffer area of tenant of other releasable data pages The data page of 50MB is put, and the data page of the 50MB released is added in the destination buffer.

Optionally, if the following buffer area hit rate of the target tenant is less than minimal buffering area hit rate, institute It states target tenant and belongs to the first priority tenant, if the following buffer area hit rate of the target tenant is greater than the minimal buffering Area's hit rate and being less than expectation buffer area hit rate, then the target tenant belongs to the second priority tenant, and described first Priority tenant is compared to the permission for the second priority tenant with data page in preferential addition buffer area.In the free time It can be preferably tenant's offer data page of high demand, guarantee multi-tenant is whole as far as possible in buffer area when data page deficiency Body performance.For example, the corresponding following buffer area hit rate of tenant T1 is less than the corresponding minimal buffering area hit rate of tenant T1, tenant The corresponding following buffer area hit rate of T2 is greater than the corresponding minimal buffering area hit rate of tenant T2 and is less than the tenant T2 corresponding phase Hope buffer area hit rate, then the server priority is the data page in tenant T1 addition buffer area, then slow for tenant T2 addition The data page in area is rushed, the preferential buffer requirements for ensureing high priority tenant.

As it can be seen that in the method depicted in fig. 2, the following buffer area of the target tenant in multi-tenant obtained according to prediction Hit rate and the preset buffer area hit rate of the target tenant, dynamically to adjust the corresponding mesh of target tenant in multi-tenant The capacity of buffer area is marked, not only can be to avoid the access performance for influencing tenant because of out of buffers, but also it can be to avoid because of buffer area It is excessive and result in waste of resources.

For the ease of better implementing the above-mentioned buffer management method towards multi-tenant of the embodiment of the present invention, the present invention It additionally provides for realizing the server of the above method is implemented.

Fig. 3 is referred to, is a kind of structural schematic diagram of server provided in an embodiment of the present invention.Service as shown in Figure 3 Device 30 can be showed in the form of general purpose computing device.The component of server 30 can include but is not limited to: one or more Processor or processing unit 301, connect different system components at memory 302 (including processor 301 and memory 302) Bus 303.Wherein,

Bus 303 indicates one of a few class bus structures or a variety of, including memory bus or Memory Controller, Peripheral bus, graphics acceleration port, processor or the local bus using any bus structures in a variety of bus structures.It lifts For example, these architectures include but is not limited to industry standard architecture (Industry Standard Architecture, ISA) bus, microchannel architecture (Micro Channel Architecture, MAC) bus, enhancing Type isa bus, Video Electronics Standards Association (Video Electronics Standards Association, VESA) local Bus and peripheral component interconnection (Peripheral Component Interconnect, PCI) bus.

Server 30 typically comprises a variety of computer system readable media, these media can be and any can be serviced The usable medium that device 30 accesses, including volatile and non-volatile media, moveable and immovable medium.

Memory 302 may include the computer system readable media of form of volatile memory, such as arbitrary access is deposited Reservoir (Random Access Memory, RAM) 3021 and/or cache memory 3022.Server 30 can be further Including other removable/nonremovable, volatile/non-volatile computer system storage mediums.Only as an example, storage system System 3023 can be used for reading and writing immovable, non-volatile magnetic media (Fig. 3 do not show, commonly referred to as " hard disk drive "). Although being not shown in Fig. 3, the disc driver for reading and writing to removable non-volatile magnetic disk (such as " floppy disk ") can be provided, And the CD drive to removable anonvolatile optical disk (such as CD-ROM, DVD-ROM or other optical mediums) read-write.In In the case of these, each driver can be connected by one or more data media interfaces with bus 303.Memory 302 For storing data, application data, the result data that processor executes or processor are specifically included and executes the data needed, Memory 302 may include at least one program product, which has one group of (for example, at least one) program module, this A little program modules are configured to perform the buffer management method towards multi-tenant described in various embodiments of the present invention.

Server 30 can also have the program/utility 3024 of one group of (at least one) program module 3025, can be with Be stored in such as memory 302, such program module 3025 includes but is not limited to: operating system, one or more answers With program, other program modules and program data, it may include network rings in each of these examples or certain combination The realization in border.Program module 3025 usually executes the buffer management method described in the invention towards multi-tenant.

Server 30 can also be with one or more external equipments 32 (such as keyboard, mouse, voice-input device, touch Output equipments such as the input equipments such as input equipment and display 31, loudspeaker, printer etc.) connection.Server 30 can also wrap Containing permissible for example, by local area network (LAN), wide area network (WAN) and/or public network (such as internet etc.) distributed computing Wireless network in environment is come the communication interface 304 that is communicated with other computer equipments 34.Other computer equipments 34 can To include server, client computer (the including but not limited to platform for executing application program associated with data access and directory service The mobile terminal devices such as formula computer, laptop, tablet computer, mobile phone) etc., which is specifically as follows each tenant Client computer can be used for example, by network access servers such as local area network, internets in corresponding client computer, multiple tenants 30, thus the tenant data needed for obtaining.Communication interface 304 is an example of communication media.Communication media is usually by such as Computer readable instructions, data structure, program module or other in the modulated message signals such as carrier wave or other transmission mechanisms Data embody, and including any information transmitting medium.Term " modulated message signal " refers to one or more feature The signal for being set or changing in a manner of encoded information in the signal.As an example, not a limit, communication media includes wired The connection of medium, such as cable network or direct line and wireless medium, such as acoustics, RF (Radio Frequency, radio frequency), Infrared ray and other wireless mediums.Term used herein computer-readable medium includes storage medium and communication media two Person.

Server 30 can also connect at least one hardware data library facilities 33 (including database 1- number by system bus According to library N), the data of multiple tenants can be stored on database facility, and the data of each tenant are mutually isolated, it is ensured that multiple rents The safety of user data.For server 30 when receiving request of the tenant by client computer access target data, server 30 is first First check whether the buffer area of operating system has been read out come the target data for determining that the tenant wants to inquire, if the mesh Mark data be previously removed, then server 30, which will take out the target data and be provided it to from buffer area, makes access The tenant of request, if the target data was previously removed not yet, this is simultaneously taken out in 30 access associated data library of server Target data is supplied to the tenant for making access request.

It will be understood by those skilled in the art that the structure of server 30 shown in Fig. 3 does not constitute the limit to server It is fixed, it may include perhaps combining certain components or different component layouts than illustrating more or fewer components, it should be bright It is white, although not shown in the drawings, other hardware and/or software module can be used in conjunction with server 30, including but not limited to: micro- generation Code, device driver, redundant processing unit, external disk drive array, disk array (Redundant Arrays of Independent Disks, RAID) system, tape drive and data backup storage system etc..

Processor 301 is used to read the buffer management code towards multi-tenant of the storage of memory 304, and executes Following operation:

Processor 301 is according to the history buffer hit rate of target tenant in default historical time section, according to preset rules Predict the following buffer area hit rate of the target tenant, the target tenant is one in the multi-tenant；

The following buffer area hit rate of processor 301 and the preset expectation buffer area hit rate of the target tenant Size；

If it is described future buffer area hit rate be greater than expectation buffer area hit rate, processor 301 according to it is described not The difference for coming buffer area hit rate Yu target tenant preset first pre-set buffer area hit rate, according to the first preset algorithm The first capacity is calculated, and discharges data page from the corresponding destination buffer of the target tenant according to first capacity； Wherein first pre-set buffer area's hit rate is less than the following buffer area hit rate；

If the future buffer area hit rate is less than expectation buffer area hit rate, processor 301 is according to the mesh The difference for marking tenant's preset second pre-set buffer area's hit rate and the following buffer area hit rate, according to the second preset algorithm The second capacity is calculated, and is that data page is added in the destination buffer according to second capacity；Wherein described second is default Buffer area hit rate is greater than the following buffer area hit rate.

Server 30 can be preset according to the buffer area hit rate demand in target tenant's future and the target tenant The size relation of buffer area hit rate demand dynamically adjusts the capacity of the corresponding destination buffer of the target tenant.

Optionally, processor 301 is default according to the following buffer area hit rate and the target tenant preset first The difference of buffer area hit rate calculates the first capacity according to the first preset algorithm, comprising:

Processor 301 is ordered the expectation buffer area hit rate as the target tenant preset first pre-set buffer area Middle rate, and held according to the difference and preset buffer area of the following buffer area hit rate and the expectation buffer area hit rate Amount, calculates the first capacity according to the first preset algorithm.

Server 30 can be in the case where meeting the expectation buffer area hit rate demand of the target tenant, using described The following buffer area hit rate of target tenant and the difference of the expectation buffer area hit rate calculate, and obtain the target and rent The capacity of family data page more than needed avoids the wasting of resources so that the data page that the target tenant has more than needed be discharged.

Optionally, processor 301 is according to target tenant preset second pre-set buffer area's hit rate and the future The difference of buffer area hit rate calculates the second capacity according to the second preset algorithm, comprising:

Processor 301 is ordered the expectation buffer area hit rate as the target tenant preset second pre-set buffer area Middle rate, and held according to the difference and preset buffer area of the expectation buffer area hit rate and the following buffer area hit rate Amount, calculates the second capacity according to the second preset algorithm.

For the expectation buffer area hit rate demand for ensureing the target tenant, server 30 utilizes the phase of the target tenant It hopes that buffer area hit rate and the difference of the following buffer area hit rate are calculated, it is required at present to obtain the target tenant The capacity of data page ensures the access performance demand of the target tenant to add data page for the target tenant.

Processor 301 judges whether the following buffer area hit rate is greater than the preset maximum buffer of target tenant Hit rate, the maximum buffer hit rate are greater than expectation buffer area hit rate；

If so, processor 301 is preset the maximum buffer hit rate as the target tenant preset first Buffer area hit rate, and according to the difference of the following buffer area hit rate and maximum preset buffer area hit rate and pre- If buffer pool size, calculate the first capacity according to the first preset algorithm.

Server 30 can be under the premise of meeting the maximum buffer hit rate demand of the target tenant, using described The difference of the following buffer area hit rate and the maximum buffer hit rate is calculated, and obtains what the destination buffer was had more than needed The capacity of data page avoids the wasting of resources so that the data page that the target tenant has more than needed be discharged.

Optionally, processor 301 is discharged from the corresponding destination buffer of the target tenant according to first capacity Data page, comprising:

Processor 301 discharges data page from the corresponding destination buffer of the target tenant according to first capacity, And the data page of first capacity released is added in freebuf, the data page in the freebuf is used The tenant for needing to add data page in for the multi-tenant uses.

The data page released from the buffer area of the target tenant can be added to freebuf by server 30 In, it can not only ensure the access performance demand of the target tenant, but also the buffer resource that the target tenant can be had more than needed Release, with supply other in the multi-tenant need to increase buffer pool size tenant it is (including the target tenant, described The tenant that other tenants and the server in multi-tenant newly increase) or server in other business use, improve The overall performance of multi-tenant and the resource utilization of buffer area.

If it is not, then processor 301 is preset the expectation buffer area hit rate as the target tenant preset first Buffer area hit rate, and according to the following buffer area hit rate and the difference for it is expected buffer area hit rate and preset Buffer pool size calculates the first capacity according to the first preset algorithm.

Processor 301 discharges data page from the corresponding destination buffer of the target tenant according to first capacity, The tenant that the data page released is used to need to add for other in the multi-tenant data page uses, first capacity Need to add tenant's data page capacity to be added of data page less than or equal to described other.

Server 30 can be when other tenants have data page demand in the multi-tenant, will be from the slow of the target tenant It rushes the data page released in area and distributes to other tenants use, can both ensure the access performance demand of the target tenant, The buffer resource that the target tenant has more than needed can be released again, need to add number to supply other in the multi-tenant It is used according to the tenant of page, improves the overall performance of multi-tenant and the resource utilization of buffer area.

Processor 301 judges whether the following buffer area hit rate is less than the preset minimal buffering area of the target tenant Hit rate, minimal buffering area hit rate are less than expectation buffer area hit rate；

If so, processor 301 is preset minimal buffering area hit rate as the target tenant preset second Buffer area hit rate, and according to the difference of minimal buffering area hit rate and the following buffer area hit rate and preset Buffer pool size calculates the second capacity according to the second preset algorithm.

Server 30 is the minimal buffering area hit rate demand for ensureing the target tenant, most using the target tenant Minibuffer area hit rate and the difference of the following buffer area hit rate are calculated, and show that the target tenant at least needs to add Add the data page of the second capacity, to add data page for the target tenant, ensures that the access performance of the target tenant needs It asks.

If it is not, then processor 301 is preset the expectation buffer area hit rate as the target tenant preset second Buffer area hit rate, and according to the difference of the expectation buffer area hit rate and the future buffer area hit rate and preset Buffer pool size calculates the second capacity according to the second preset algorithm.

Server 30 is the expectation buffer area hit rate demand for ensureing the target tenant, utilizes the phase of the target tenant It hopes that buffer area hit rate and the difference of the following buffer area hit rate are calculated, it is required at present to obtain the target tenant The capacity of data page ensures the access performance demand of the target tenant to add data page for the target tenant.

Optionally, first preset algorithm are as follows: C1=(A-B) * D, wherein C1 is first capacity, A be it is described not Carry out buffer area hit rate, B is first pre-set buffer area hit rate, and D is the preset buffer pool size.

Optionally, first preset algorithm are as follows: C1=min ((A-B) * D, E), wherein C1 is first capacity, A For the following buffer area hit rate, B is first pre-set buffer area hit rate, and D is the preset buffer pool size, E For the data page capacity to be added of other tenants in the multi-tenant.

Optionally, second preset algorithm are as follows: C2=min ((F-A) * D, G), wherein C2 is second capacity, F For second pre-set buffer area hit rate, A is the following buffer area hit rate, and D is the preset buffer pool size, G For the sum of the data page capacity in other tenants in the multi-tenant data page capacity and freebuf to be released.

Optionally, the preset buffer pool size include: the preset minimal buffering area capacity of the target tenant, it is described The preset expectation buffer pool size of target tenant, the preset maximum buffer capacity of the target tenant and the destination buffer At least one of current capacity.

Optionally, processor 301 is executed according to the history buffer hit rate for presetting target tenant in historical time section, is pressed Predict that the operation of the following buffer area hit rate of the target tenant is triggered by least one following event according to preset rules:

Predetermined period reaches；Preset time point reaches；Buffer area hit rate is lower than the rent of pre-set buffer area hit rate threshold value The number at family is greater than or equal to predetermined number threshold value.When meeting trigger event, server can understand the target in time and rent The buffer area hit rate situation at family, and then the capacity of the destination buffer is adjusted in time.

It should be noted that the function of each functional module can be according to above-mentioned in server 30 described in the embodiment of the present invention Method specific implementation in embodiment of the method shown in Fig. 2, details are not described herein again.

Fig. 4 is referred to, is the structural schematic diagram of another server provided in an embodiment of the present invention.As shown in figure 4, service Device 40 can include: predicting unit 401, comparing unit 403, the first administrative unit 405 and the second administrative unit 407, wherein

Predicting unit 401, for the history buffer hit rate according to target tenant in default historical time section, according to pre- If rule predicts the following buffer area hit rate of the target tenant, the target tenant is one in the multi-tenant；

Comparing unit 403, the following buffer area hit rate obtained for the predicting unit 401 prediction and institute State the size of the preset expectation buffer area hit rate of target tenant；

First administrative unit 405, if comparing the following buffer area hit rate for the comparing unit 403 is greater than institute Desired buffer area hit rate is stated, then according to the following buffer area hit rate and preset first pre-set buffer of the target tenant The difference of area's hit rate calculates the first capacity according to the first preset algorithm, and rents according to first capacity from the target Family discharges data page in corresponding destination buffer；Wherein first pre-set buffer area's hit rate is less than the following buffer area Hit rate；

Second administrative unit 407, if comparing the following buffer area hit rate for the comparing unit 403 is less than institute Desired buffer area hit rate is stated, then according to target tenant preset second pre-set buffer area's hit rate and the following buffering The difference of area's hit rate calculates the second capacity according to the second preset algorithm, and slow for the target according to second capacity Rush area's addition data page；Wherein second pre-set buffer area's hit rate is greater than the following buffer area hit rate.

Optionally, first administrative unit 405 is specifically used for:

Optionally, second administrative unit 407 is specifically used for:

Optionally, first administrative unit 405 is specifically used for:

In the embodiment of the present invention, server 40 can be according to the buffer area hit rate demand in target tenant's future and described The size relation of the preset buffer area hit rate demand of target tenant dynamically adjusts the corresponding destination buffer of the target tenant Capacity.

It should be noted that the function of each functional unit can be according to above-mentioned in server 40 described in the embodiment of the present invention Method specific implementation in embodiment of the method shown in Fig. 2, details are not described herein again.

Fig. 5 is referred to, is the structural schematic diagram of another server provided in an embodiment of the present invention.Wherein, shown in fig. 5 Server 40 is that server 40 as shown in Figure 4 optimizes.Compared with Fig. 4, server 40 shown in fig. 5 equally packet Include predicting unit 401, comparing unit 403, the first administrative unit 405 and the second administrative unit 407, and the first administrative unit 405 It specifically includes: the first judging unit 4051 and the first computing unit 4053, wherein

First judging unit 4051 is greater than if comparing the following buffer area hit rate for the comparing unit 403 Expectation buffer area hit rate, then judge whether the following buffer area hit rate is greater than the preset maximum of target tenant Buffer area hit rate, the maximum buffer hit rate are greater than expectation buffer area hit rate；

First computing unit 4053, if judging the following buffer area hit rate for first judging unit 4051 Greater than the preset maximum buffer hit rate of the target tenant, then rented using the maximum buffer hit rate as the target Family preset first pre-set buffer area hit rate, and ordered according to the following buffer area hit rate and the maximum preset buffer area The difference of middle rate and preset buffer pool size calculate the first capacity according to the first preset algorithm, and according to described first Capacity discharges data page from the corresponding destination buffer of the target tenant.

Fig. 6 is referred to, is the structural schematic diagram of another server provided in an embodiment of the present invention.Wherein, shown in fig. 6 Server 40 is that server 40 as shown in Figure 4 optimizes.Compared with Fig. 4, server 40 shown in fig. 6 equally packet Include predicting unit 401, comparing unit 403, the first administrative unit 405 and the second administrative unit 407, and the first administrative unit 405 It specifically includes: second judgment unit 4055 and the second computing unit 4057, wherein

Second judgment unit 4055 is greater than if comparing the following buffer area hit rate for the comparing unit 403 Expectation buffer area hit rate, then judge whether the following buffer area hit rate is greater than the preset maximum of target tenant Buffer area hit rate, the maximum buffer hit rate are greater than expectation buffer area hit rate；

Second computing unit 4057, if judging the following buffer area hit rate for the second judgment unit 4055 Less than or equal to the preset maximum buffer hit rate of the target tenant, then using expectation buffer area hit rate as described in Target tenant preset first pre-set buffer area hit rate, and according to the following buffer area hit rate and the expectation buffer area The difference of hit rate and preset buffer pool size calculate the first capacity according to the first preset algorithm, and according to described One capacity discharges data page from the corresponding destination buffer of the target tenant.

Fig. 7 is referred to, is the structural schematic diagram of another server provided in an embodiment of the present invention.Wherein, shown in Fig. 7 Server 40 is that server 40 as shown in Figure 4 optimizes.Compared with Fig. 4, server 40 shown in Fig. 7 equally packet Include predicting unit 401, comparing unit 403, the first administrative unit 405 and the second administrative unit 407, and the second administrative unit 407 It specifically includes: third judging unit 4071 and third computing unit 4073, wherein

Third judging unit 4071 is less than if comparing the following buffer area hit rate for the comparing unit 403 Expectation buffer area hit rate, then judge whether the following buffer area hit rate is less than the preset minimum of target tenant Buffer area hit rate, minimal buffering area hit rate are less than expectation buffer area hit rate；

Third computing unit 4073, if judging the following buffer area hit rate for the third judging unit 4071 Less than the preset minimal buffering area hit rate of the target tenant, then rented minimal buffering area hit rate as the target Family preset second pre-set buffer area hit rate, and according to minimal buffering area hit rate and the following buffer area hit rate Difference and preset buffer pool size, calculate the second capacity according to the second preset algorithm, and according to second capacity Data page is added for the destination buffer.

Fig. 8 is referred to, is the structural schematic diagram of another server provided in an embodiment of the present invention.Wherein, shown in Fig. 8 Server 40 is that server 40 as shown in Figure 4 optimizes.Compared with Fig. 4, server 40 shown in Fig. 8 equally packet Include predicting unit 401, comparing unit 403, the first administrative unit 405 and the second administrative unit 407, and the second administrative unit 407 It specifically includes: the 4th judging unit 4075 and the 4th computing unit 4077, wherein

4th judging unit 4075 is less than if comparing the following buffer area hit rate for the comparing unit 403 Expectation buffer area hit rate, then judge whether the following buffer area hit rate is less than the preset minimum of target tenant Buffer area hit rate, minimal buffering area hit rate are less than expectation buffer area hit rate；

4th computing unit 4077, if judging the following buffer area hit rate for the 4th judging unit 4075 More than or equal to the preset minimal buffering area hit rate of the target tenant, then using expectation buffer area hit rate as described in Target tenant preset second pre-set buffer area hit rate, and according to the expectation buffer area hit rate and the following buffer area The difference of hit rate and preset buffer pool size calculate the second capacity according to the second preset algorithm, and according to described Two capacity are that data page is added in the destination buffer.

Optionally, above-mentioned first administrative unit 405 can also include the first judging unit 4051, the first computing unit simultaneously 4053, second judgment unit 4055 and the second computing unit 4057, the specific implementation of each unit are referred in above scheme The description of same unit, details are not described herein again.

Optionally, above-mentioned second administrative unit 407 can also include third judging unit 4071, third computing unit simultaneously 4073, the specific implementation of the 4th judging unit 4075 and the 4th computing unit 4077, each unit is referred in above scheme The description of same unit, details are not described herein again.

In conclusion implementing the embodiment of the present invention, the following of the target tenant in multi-tenant obtained according to prediction is buffered Area's hit rate and the preset buffer area hit rate of the target tenant, it is corresponding dynamically to adjust the target tenant in multi-tenant The capacity of destination buffer, not only can to avoid the access performance for influencing tenant because of out of buffers, but also can to avoid because buffering Area is excessive and results in waste of resources.

Further, its in multi-tenant will be supplied from the data page that the corresponding destination buffer of the target tenant releases The tenant that he needs to add data page uses, and the overall performance of multi-tenant and the resource utilization of buffer area can be improved.

Those of ordinary skill in the art will appreciate that realizing all or part of the process in above-described embodiment method, being can be with Relevant hardware is instructed to complete by computer program, the program can be stored in a computer-readable storage medium In, the program is when being executed, it may include such as the process of the embodiment of above-mentioned each method.Wherein, the storage medium can be magnetic Dish, CD, read-only memory (Read-Only Memory, ROM) or random access memory etc..

The above disclosure is only the preferred embodiments of the present invention, cannot limit the right model of the present invention with this certainly It encloses, therefore equivalent changes made in accordance with the claims of the present invention, is still within the scope of the present invention.

Claims

1. a kind of buffer management method towards multi-tenant characterized by comprising

According to the history buffer hit rate of target tenant in default historical time section, predict that the target is rented according to preset rules The following buffer area hit rate at family, the target tenant are one in the multi-tenant；

Compare the size of the following buffer area hit rate and the preset expectation buffer area hit rate of the target tenant；

If the future buffer area hit rate is greater than expectation buffer area hit rate, according to the following buffer area hit rate With the difference of target tenant preset first pre-set buffer area hit rate, the first appearance is calculated according to the first preset algorithm Amount, and data page is discharged from the corresponding destination buffer of the target tenant according to first capacity；Wherein described first Pre-set buffer area hit rate is less than the following buffer area hit rate；

If the future buffer area hit rate is less than expectation buffer area hit rate, according to the target tenant preset the The difference of two pre-set buffer area hit rates and the following buffer area hit rate, calculates the second appearance according to the second preset algorithm Amount, and be that data page is added in the destination buffer according to second capacity；Wherein second pre-set buffer area hit rate Greater than the following buffer area hit rate.

2. the method according to claim 1, wherein described according to the following buffer area hit rate and the mesh The difference for marking tenant's preset first pre-set buffer area hit rate, calculates the first capacity according to the first preset algorithm, comprising:

Using the expectation buffer area hit rate as target tenant preset first pre-set buffer area hit rate, and according to institute The following buffer area hit rate and the difference for it is expected buffer area hit rate and preset buffer pool size are stated, it is pre- according to first Imputation method calculates the first capacity.

3. the method according to claim 1, wherein described default slow according to the target tenant preset second The difference for rushing area's hit rate and the following buffer area hit rate, calculates the second capacity according to the second preset algorithm, comprising:

Using the expectation buffer area hit rate as target tenant preset second pre-set buffer area hit rate, and according to institute State desired buffer area hit rate and the following buffer area hit rate difference and preset buffer pool size, it is pre- according to second Imputation method calculates the second capacity.

4. the method according to claim 1, wherein described according to the following buffer area hit rate and the mesh The difference for marking tenant's preset first pre-set buffer area hit rate, calculates the first capacity according to the first preset algorithm, comprising:

Judge whether the following buffer area hit rate is greater than the preset maximum buffer hit rate of the target tenant, it is described most Big buffer area hit rate is greater than expectation buffer area hit rate；

If so, using the maximum buffer hit rate as target tenant preset first pre-set buffer area hit rate, And held according to the difference and preset buffer area of the following buffer area hit rate and maximum preset buffer area hit rate Amount, calculates the first capacity according to the first preset algorithm.

5. method according to claim 1,2 or 4, which is characterized in that it is described according to first capacity from the target Tenant discharges data page in corresponding destination buffer, comprising:

Data page, and the institute that will be released are discharged from the corresponding destination buffer of the target tenant according to first capacity The data page for stating the first capacity is added in freebuf, and the data page in the freebuf is used to supply the multi-tenant The middle tenant for needing to add data page uses.

6. the method according to claim 1, wherein described according to the following buffer area hit rate and the mesh The difference for marking tenant's preset first pre-set buffer area hit rate, calculates the first capacity according to the first preset algorithm, comprising:

If it is not, then using the expectation buffer area hit rate as target tenant preset first pre-set buffer area hit rate, And according to the difference and preset buffer pool size of the following buffer area hit rate and the expectation buffer area hit rate, press The first capacity is calculated according to the first preset algorithm.

7. the method according to claim 2 or 6, which is characterized in that described to be rented according to first capacity from the target Family discharges data page in corresponding destination buffer, comprising:

Data page, the number released are discharged from the corresponding destination buffer of the target tenant according to first capacity It is used according to the tenant that page is used to need to add for other in the multi-tenant data page, first capacity is less than or equal to described Other need to add tenant's data page capacity to be added of data page.

8. the method according to claim 1, wherein described default slow according to the target tenant preset second The difference for rushing area's hit rate and the following buffer area hit rate, calculates the second capacity according to the second preset algorithm, comprising:

Judge whether the following buffer area hit rate is less than the preset minimal buffering area hit rate of the target tenant, it is described most Minibuffer area hit rate is less than expectation buffer area hit rate；

If so, using minimal buffering area hit rate as target tenant preset second pre-set buffer area hit rate, And according to the difference and preset buffer pool size of minimal buffering area hit rate and the following buffer area hit rate, press The second capacity is calculated according to the second preset algorithm.

9. the method according to claim 1, wherein described default slow according to the target tenant preset second The difference for rushing area's hit rate and the following buffer area hit rate, calculates the second capacity according to the second preset algorithm, comprising:

If it is not, then using the expectation buffer area hit rate as target tenant preset second pre-set buffer area hit rate, And according to the difference and preset buffer pool size of the expectation buffer area hit rate and the following buffer area hit rate, press The second capacity is calculated according to the second preset algorithm.

10. according to claim 1, method described in 2,4 or 6, which is characterized in that first preset algorithm are as follows: C1=(A- B) * D, wherein C1 is first capacity, and A is the following buffer area hit rate, and B is the hit of first pre-set buffer area Rate, D are the preset buffer pool size.

11. according to claim 1, method described in 2 or 6, which is characterized in that first preset algorithm are as follows: C1=min ((A- B) * D, E), wherein C1 is first capacity, and A is the following buffer area hit rate, and B is first pre-set buffer area life Middle rate, D are the preset buffer pool size, and E is the data page capacity that other tenants are to be added in the multi-tenant.

12. according to claim 1, method described in 3,8 or 9, which is characterized in that second preset algorithm are as follows: C2=min ((F-A) * D, G), wherein C2 is second capacity, and F is second pre-set buffer area hit rate, and A is the following buffering Area's hit rate, D are the preset buffer pool size, G be in the multi-tenant other tenants data page capacity to be released and The sum of data page capacity in freebuf.

13. according to claim 2,3,4,6,8 or 9 described in any item methods, which is characterized in that the preset buffer area holds Amount includes: the preset minimal buffering area capacity of the target tenant, the preset expectation buffer pool size of the target tenant, described Current at least one of the capacity of the preset maximum buffer capacity of target tenant and the destination buffer.

14. a kind of server characterized by comprising

Predicting unit, for the history buffer hit rate according to target tenant in default historical time section, according to preset rules Predict the following buffer area hit rate of the target tenant, the target tenant is one in multi-tenant；

Comparing unit, the following buffer area hit rate predicted for the predicting unit and the target tenant The size of preset expectation buffer area hit rate；

First administrative unit buffers if comparing the following buffer area hit rate for the comparing unit greater than the expectation Area's hit rate, then according to the following buffer area hit rate and target tenant preset first pre-set buffer area hit rate Difference calculates the first capacity according to the first preset algorithm, and according to first capacity from the corresponding mesh of the target tenant Mark discharges data page in buffer area；Wherein first pre-set buffer area's hit rate is less than the following buffer area hit rate；

Second administrative unit buffers if comparing the following buffer area hit rate for the comparing unit less than the expectation Area's hit rate, then according to target tenant preset second pre-set buffer area's hit rate and the following buffer area hit rate Difference calculates the second capacity according to the second preset algorithm, and is that number is added in the destination buffer according to second capacity According to page；Wherein second pre-set buffer area's hit rate is greater than the following buffer area hit rate.

15. server according to claim 14, which is characterized in that first administrative unit is specifically used for:

16. server according to claim 14, which is characterized in that second administrative unit is specifically used for:

17. server according to claim 14, which is characterized in that first administrative unit includes:

First judging unit buffers if comparing the following buffer area hit rate for the comparing unit greater than the expectation Area's hit rate, then judge whether the following buffer area hit rate is greater than the preset maximum buffer hit of the target tenant Rate, the maximum buffer hit rate are greater than expectation buffer area hit rate；

First computing unit, if judging that the following buffer area hit rate is greater than the target for first judging unit The preset maximum buffer hit rate of tenant, then using the maximum buffer hit rate as the target tenant preset first Pre-set buffer area hit rate, and according to the difference of the following buffer area hit rate and maximum preset buffer area hit rate with And preset buffer pool size, the first capacity is calculated according to the first preset algorithm, and according to first capacity from the mesh Mark tenant discharges data page in corresponding destination buffer.

18. server described in 4,15 or 17 according to claim 1, which is characterized in that first administrative unit is specifically used for:

19. server according to claim 14, which is characterized in that first administrative unit includes:

Second judgment unit buffers if comparing the following buffer area hit rate for the comparing unit greater than the expectation Area's hit rate, then judge whether the following buffer area hit rate is greater than the preset maximum buffer hit of the target tenant Rate, the maximum buffer hit rate are greater than expectation buffer area hit rate；

Second computing unit, if judging that the following buffer area hit rate is less than or equal to institute for the second judgment unit The preset maximum buffer hit rate of target tenant is stated, then is preset using the expectation buffer area hit rate as the target tenant The first pre-set buffer area hit rate, and according to the difference of the following buffer area hit rate and the expectation buffer area hit rate And preset buffer pool size, the first capacity is calculated according to the first preset algorithm, and according to first capacity from described Target tenant discharges data page in corresponding destination buffer.

20. server described in 5 or 19 according to claim 1, which is characterized in that first administrative unit is specifically used for:

21. server according to claim 14, which is characterized in that second administrative unit includes:

Third judging unit buffers if comparing the following buffer area hit rate for the comparing unit less than the expectation Area's hit rate, then judge whether the following buffer area hit rate is less than the preset minimal buffering area hit of the target tenant Rate, minimal buffering area hit rate are less than expectation buffer area hit rate；

Third computing unit, if judging that the following buffer area hit rate is less than the target for the third judging unit The preset minimal buffering area hit rate of tenant, then using minimal buffering area hit rate as the target tenant preset second Pre-set buffer area hit rate, and according to the difference of minimal buffering area hit rate and the following buffer area hit rate and pre- If buffer pool size, calculate the second capacity according to the second preset algorithm, and slow for the target according to second capacity Rush area's addition data page.

22. server according to claim 14, which is characterized in that second administrative unit includes:

4th judging unit buffers if comparing the following buffer area hit rate for the comparing unit less than the expectation Area's hit rate, then judge whether the following buffer area hit rate is less than the preset minimal buffering area hit of the target tenant Rate, minimal buffering area hit rate are less than expectation buffer area hit rate；

4th computing unit, if judging that the following buffer area hit rate is greater than or equal to institute for the 4th judging unit The preset minimal buffering area hit rate of target tenant is stated, then is preset using the expectation buffer area hit rate as the target tenant The second pre-set buffer area hit rate, and according to the difference of the expectation buffer area hit rate and the following buffer area hit rate And preset buffer pool size, the second capacity is calculated according to the second preset algorithm, and be described according to second capacity Add data page in destination buffer.

23. server described in 4,15,17 or 19 according to claim 1, which is characterized in that first preset algorithm are as follows: C1 =(A-B) * D, wherein C1 is first capacity, and A is the following buffer area hit rate, and B is first pre-set buffer area Hit rate, D are the preset buffer pool size.

24. server described in 4,15 or 19 according to claim 1, which is characterized in that first preset algorithm are as follows: C1= Min ((A-B) * D, E), wherein C1 is first capacity, and A is the following buffer area hit rate, and B is described first default Buffer area hit rate, D are the preset buffer pool size, and E is that the data page that other tenants are to be added in the multi-tenant holds Amount.

25. server described in 4,16,21 or 22 according to claim 1, which is characterized in that second preset algorithm are as follows: C2 =min ((F-A) * D, G), wherein C2 is second capacity, and F is second pre-set buffer area hit rate, A be it is described not Carry out buffer area hit rate, D is the preset buffer pool size, and G is the data page that other tenants are to be released in the multi-tenant The sum of data page capacity in capacity and freebuf.

26. the described in any item servers in 5,16,17,19,21 or 22 according to claim 1, which is characterized in that described preset Buffer pool size includes: the preset minimal buffering area capacity of the target tenant, the preset expectation buffer area of the target tenant At least one of capacity, the current capacity of the preset maximum buffer capacity of the target tenant and the destination buffer.

27. a kind of server, which is characterized in that including processor and memory, wherein the memory is for storing towards more The buffer management code of tenant, the processor are used to that the program code perform claim of the memory storage to be called to want The buffer management method towards multi-tenant of asking 1-13 described in any item.