CN103530253B

CN103530253B - Cluster many overall situations Buffer Pool system, Centroid, calculating node and management method

Info

Publication number: CN103530253B
Application number: CN201310462330.8A
Authority: CN
Inventors: 程广卫
Original assignee: Huawei Technologies Co Ltd
Current assignee: Huawei Technologies Co Ltd
Priority date: 2013-09-30
Filing date: 2013-09-30
Publication date: 2016-08-17
Anticipated expiration: 2033-09-30
Also published as: CN103530253A

Abstract

Embodiments providing a kind of cluster many overall situations Buffer Pool system, Centroid, calculating node and management method, this system includes: overall situation Buffer Pool, for storing the globally shared buffered data of this system；Calculate node, for sending page request to Centroid；Wherein, this page request carries the node serial number calculating node and the page number calculating the memory pages that node application accesses；Centroid, can be according to the page number of the node serial number and this memory pages calculating node, there is provided this memory pages storage positional information calculation node in overall situation Buffer Pool to be additionally operable to the storage positional information in overall situation Buffer Pool of this memory pages according to Centroid offer to calculating node, this memory pages is conducted interviews.In the embodiment of the present invention, the resource of overall situation Buffer Pool is managed distributing and the Centroid possibility that make overall situation Buffer Pool possess flexible expansion separate with overall situation Buffer Pool by Centroid according to the request calculating node.

Description

Cluster many overall situations Buffer Pool system, Centroid, calculating node and management method

Technical field

The present embodiments relate to computer realm, and delay more particularly, to a kind of many overall situations of cluster Rush cell system, Centroid, calculating node and management method.

Background technology

In the cluster that there is Centroid, especially there are the data-base cluster of Centroid, Centroid The assembly of upper existence referred to as overall situation Buffer Pool.Overall situation Buffer Pool serves all nodes in cluster, and is These nodes provide data buffer storage.Especially support remote direct memory access (RDMA) low time Prolonging on the network of high bandwidth, cluster interior nodes is higher than local disk to the access efficiency of overall situation Buffer Pool, and And overall situation Buffer Pool achieves and shares between all nodes, provide greatly for cluster interior nodes exchange data Facility.

But, owing to overall situation Buffer Pool is on Centroid, its extension is limited to the internal memory of Centroid and holds Amount, it is impossible to flexible expansion.

Summary of the invention

The embodiment of the present invention provide a kind of cluster many overall situations Buffer Pool system, Centroid, calculating node and Management method, it is possible to reduce overall situation Buffer Pool and the Centroid degree of coupling, it is simple to overall situation Buffer Pool flexible Extension.

First aspect, it is provided that a kind of cluster multiple buffer pool system, during this cluster multiple buffer pool system includes Heart node, overall situation Buffer Pool and calculating node.Wherein, this overall situation Buffer Pool is for storing the complete of this system Office's shared buffer data, this overall situation Buffer Pool includes the buffer joint independent of this Centroid；This calculating Node is for sending page request to this Centroid, and this page request is carried the node of this calculating node and compiled Number and the page number of memory pages that accesses of this calculating node application；This Centroid is based on according to this The node serial number of operator node and the page number of this memory pages, provide this memory pages to this calculating node Storage positional information in this overall situation Buffer Pool；This calculating node is additionally operable to according to this memory pages at this Storage positional information in overall situation Buffer Pool, conducts interviews to this memory pages.

In conjunction with first aspect, in the implementation that the first is possible, it is implemented as: this Centroid It is additionally operable to manage the buffer joint of this overall situation Buffer Pool.

In conjunction with the first possible implementation of first aspect, in the implementation that the second is possible, At the buffer joint for managing this overall situation Buffer Pool, this Centroid is implemented as: receives first and delays Rush the registration information of node, this registration information carry this first buffer joint node serial number and Memory source information, wherein, this first buffer joint is the buffer joint outside this overall situation Buffer Pool；According to It is complete that the memory source of this first buffer joint is registered to this by the memory source information of this first buffer joint In office's Buffer Pool.

In conjunction with the implementation that the second of first aspect is possible, in the implementation that the third is possible, Be implemented as: this memory source information include the memory address of this first buffer joint, internal memory length and Remote direct memory accesses RDMA access key KEY.

In conjunction with the first possible implementation of first aspect, in the 4th kind of possible implementation, It is implemented as: at the buffer joint for managing this overall situation Buffer Pool, this Centroid is implemented as: Receiving the de-registration request information that the second buffer joint sends, this de-registration request information carries this second buffering joint The node serial number of point, wherein, this second buffer joint is the buffer joint in this overall situation Buffer Pool；According to This de-registration request information is by the Data Migration of this second buffer joint to this overall situation Buffer Pool, this second delays Rush in the buffer joint beyond node；The memory source information of this second buffer joint is buffered from this overall situation Pond is deleted.

In conjunction with the first possible implementation of first aspect, in the 5th kind of possible implementation, It is implemented as: at the buffer joint for managing this overall situation Buffer Pool, this Centroid is implemented as: If receiving the notice about the second buffer joint fault, then according to this notice by this second buffer joint Data Migration in the buffer joint beyond this second buffer joint in this overall situation Buffer Pool, and by this second The memory source information of buffer joint is deleted from this overall situation Buffer Pool, and wherein, this second buffer joint is Buffer joint in this overall situation Buffer Pool.

In conjunction with the first possible implementation of first aspect, in the 6th kind of possible implementation, It is implemented as: at the buffer joint for managing this overall situation Buffer Pool, this Centroid is implemented as: Receiving the replacement request information that the second buffer joint sends, this replacement request information carries the 3rd buffer joint Memory source information, wherein, this second buffer joint is the buffer joint in this overall situation Buffer Pool, should 3rd buffer joint is the buffer joint outside this overall situation Buffer Pool；According to this replacement request information by this second The Data Migration of buffer joint is in the 3rd buffer joint, and deletes the registration letter of this second buffer joint Breath, and according to the memory source information of the 3rd buffer joint, the memory source of the 3rd buffer joint is noted Volume is to this overall situation Buffer Pool.

In conjunction with the first possible implementation of first aspect or first aspect to the 6th of first aspect Plant any one possible implementation in possible implementation, in the 7th kind of possible implementation, It is implemented as: this overall situation Buffer Pool includes at least 2 buffer joints, the institute in this overall situation Buffer Pool There are in buffer joint at least 2 backups of the globally shared buffered data that store this overall situation Buffer Pool.

In conjunction with the 7th kind of possible implementation of first aspect, in the 8th kind of possible implementation, It is implemented as: this memory pages storage positional information in this overall situation Buffer Pool includes this memory pages The storage position of at least two buffer joint in this overall situation Buffer Pool, this calculating node is according to this internal memory This memory pages is conducted interviews and is implemented as by page storage positional information in this overall situation Buffer Pool: When this memory pages is read out accessing by this calculating node, this calculating node selects this memory pages to exist A pair this memory pages of the storage position of at least two buffer joint in this overall situation Buffer Pool is read Take access；Or when this calculating node carries out write-access to this memory pages, this calculating node selects The storage position of this memory pages at least two buffer joint in this overall situation Buffer Pool is to this page Face carries out write-access.

In conjunction with the first possible implementation of first aspect or first aspect to the 8th of first aspect Plant any one possible implementation in possible implementation, in the 9th kind of possible implementation, It is implemented as: this application page request message also carries the page byte number of this memory pages, this page The page number that byte number combines this memory pages for this Centroid determines this storage positional information.

Second aspect, it is provided that the overall Buffer Pool management method of a kind of cluster multiple buffer pool system, wherein This cluster multiple buffer pool system includes Centroid, overall situation Buffer Pool and calculates node, this overall situation Buffer Pool Including the buffer joint independent of this Centroid, the method includes: this calculating node is to this Centroid Sending page request, this page request carries node serial number and this calculating node application visit of this calculating node The page number of the memory pages asked；This Centroid is according to the node serial number of this calculating node and this internal memory The page number of the page, provides this memory pages storage position in this overall situation Buffer Pool to this calculating node Confidence ceases, and wherein, this overall situation Buffer Pool is for storing the globally shared buffering of this cluster multiple buffer pool system Data；This calculating node is according to this memory pages storage positional information in this overall situation Buffer Pool, to this Memory pages conducts interviews.

In conjunction with second aspect, in the implementation that the first is possible, the method also includes: this centromere Point manages the buffer joint of this overall situation Buffer Pool.

In conjunction with the first possible implementation of second aspect, in the implementation that the second is possible, This Centroid manages the buffer joint of this overall situation Buffer Pool and is implemented as: this Centroid receives first The registration information of buffer joint, this registration information carries the node serial number of this first buffer joint With memory source information, wherein, this first buffer joint is the buffer joint outside this overall situation Buffer Pool；Should The internal memory of this first buffer joint is provided by Centroid according to the memory source information of this first buffer joint Login is in this overall situation Buffer Pool.

In conjunction with the first possible implementation of second aspect, in the implementation that the third is possible, This Centroid manages the buffer joint of this overall situation Buffer Pool and is implemented as: this Centroid receives second The de-registration request information that buffer joint sends, this de-registration request information carries the node of this second buffer joint Numbering, wherein, this second buffer joint is the buffer joint in this overall situation Buffer Pool；This Centroid root According to this de-registration request information by the Data Migration of this second buffer joint to this overall situation Buffer Pool this second In buffer joint beyond buffer joint；This Centroid is by the memory source information of this second buffer joint Delete from this overall situation Buffer Pool.

In conjunction with the first possible implementation of second aspect, in the 4th kind of possible implementation, This Centroid manages the buffer joint of this overall situation Buffer Pool and is implemented as: if this Centroid receives To the notice about the second buffer joint fault, then this second buffering is saved by this Centroid according to this notice The Data Migration of point, and should in the buffer joint beyond this second buffer joint in this overall situation Buffer Pool The memory source information of the second buffer joint is deleted from this overall situation Buffer Pool, wherein, and this second buffering joint Point is the buffer joint in this overall situation Buffer Pool.

In conjunction with the first possible implementation of second aspect, in the 5th kind of possible implementation, This Centroid manages the buffer joint of this overall situation Buffer Pool and is implemented as: this Centroid receives second The replacement request information that buffer joint sends, this replacement request information carries the internal memory money of the 3rd buffer joint Source information, wherein, this second buffer joint is the buffer joint in this overall situation Buffer Pool, the 3rd buffering Node is the buffer joint outside this overall situation Buffer Pool；This Centroid according to this replacement request information by this The Data Migration of two buffer joints is in the 3rd buffer joint, and deletes the registration of this second buffer joint Information, and according to the memory source information of the 3rd buffer joint by the memory source of the 3rd buffer joint It is registered to this overall situation Buffer Pool.

In conjunction with the first possible implementation of second aspect or second aspect to the 5th of second aspect Plant any one possible implementation in possible implementation, in the 6th kind of possible implementation, It is implemented as: this overall situation Buffer Pool includes at least 2 buffer joints, the institute in this overall situation Buffer Pool There are in buffer joint at least 2 backups of the globally shared buffered data that store this overall situation Buffer Pool.

In conjunction with the 6th kind of possible implementation of second aspect, in the 7th kind of possible implementation, This memory pages storage positional information in this overall situation Buffer Pool includes that this memory pages delays in this overall situation Rushing the storage position of at least two buffer joint in pond, this calculating node is complete at this according to this memory pages This memory pages is conducted interviews and is implemented as by the storage positional information in office's Buffer Pool: when this calculating saves When this memory pages is read out accessing by point, this calculating node selects this memory pages to buffer in this overall situation A pair this memory pages of the storage position of at least two buffer joint in pond is read out accessing；Or When this calculating node carries out write-access to this memory pages, this calculating node selects this memory pages to exist The storage position of at least two buffer joint in this overall situation Buffer Pool carries out write and visits this memory pages Ask.

The third aspect, it is provided that the overall Buffer Pool management method of a kind of cluster multiple buffer pool system, wherein This cluster multiple buffer pool system includes Centroid, overall situation Buffer Pool and calculates node, this overall situation Buffer Pool Including the buffer joint independent of this Centroid, the method includes: this Centroid receives this calculating joint The page request that point sends, this page request carries node serial number and this calculating node Shen of this calculating node The page number of the memory pages that please access；This Centroid is according to the node serial number of this calculating node and should The page number of memory pages, obtains this memory pages storage positional information in this overall situation Buffer Pool, Wherein, this overall situation Buffer Pool is for storing the globally shared buffered data of this cluster multiple buffer pool system；Should Centroid sends this memory pages storage positional information in this overall situation Buffer Pool to this calculating node, Wherein, this memory pages storage positional information in this overall situation Buffer Pool is interior to this for this calculating node Deposit the page to conduct interviews.

In conjunction with the third aspect, in the implementation that the first is possible, the method also includes: this centromere Point manages the buffer joint of this overall situation Buffer Pool.

In conjunction with the first possible implementation of the third aspect, in the implementation that the second is possible, This Centroid manages the buffer joint of this overall situation Buffer Pool and is implemented as: this Centroid receives first The registration information of buffer joint, this registration information carries the node serial number of this first buffer joint With memory source information, wherein, this first buffer joint is the buffer joint outside this overall situation Buffer Pool；Should The internal memory of this first buffer joint is provided by Centroid according to the memory source information of this first buffer joint Login is in this overall situation Buffer Pool.

In conjunction with the first possible implementation of the third aspect, in the implementation that the third is possible, This Centroid manages the buffer joint of this overall situation Buffer Pool and is implemented as: this Centroid receives second The de-registration request information that buffer joint sends, this de-registration request information carries the node of this second buffer joint Numbering, wherein, this second buffer joint is the buffer joint in this overall situation Buffer Pool；This Centroid root According to this de-registration request information by the Data Migration of this second buffer joint to this overall situation Buffer Pool this second In buffer joint beyond buffer joint；This Centroid is by the memory source information of this second buffer joint Delete from this overall situation Buffer Pool.

In conjunction with the first possible implementation of the third aspect, in the 4th kind of possible implementation, This Centroid manages the buffer joint of this overall situation Buffer Pool and is implemented as: if this Centroid receives To the notice about the second buffer joint fault, then this second buffering is saved by this Centroid according to this notice The Data Migration of point, and should in the buffer joint beyond this second buffer joint in this overall situation Buffer Pool The memory source information of the second buffer joint is deleted from this overall situation Buffer Pool, wherein, and this second buffering joint Point is the buffer joint in this overall situation Buffer Pool.

In conjunction with the first possible implementation of the third aspect, in the 5th kind of possible implementation, This Centroid manages the buffer joint of this overall situation Buffer Pool and is implemented as: this Centroid receives second The replacement request information that buffer joint sends, this replacement request information carries the internal memory money of the 3rd buffer joint Source information, wherein, this second buffer joint is the buffer joint in this overall situation Buffer Pool, the 3rd buffering Node is the buffer joint outside this overall situation Buffer Pool；This Centroid according to this replacement request information by this The Data Migration of two buffer joints is in the 3rd buffer joint, and deletes the registration of this second buffer joint Information, and according to the memory source information of the 3rd buffer joint by the memory source of the 3rd buffer joint It is registered to this overall situation Buffer Pool.

In conjunction with the first possible implementation of the third aspect or the third aspect to the 5th of the third aspect Plant any one possible implementation in possible implementation, in the 6th kind of possible implementation, It is implemented as: this overall situation Buffer Pool includes at least 2 buffer joints, the institute in this overall situation Buffer Pool There are in buffer joint at least 2 backups of the globally shared buffered data that store this overall situation Buffer Pool, should Memory pages storage positional information in this overall situation Buffer Pool includes that this memory pages buffers in this overall situation The storage position of at least two buffer joint in pond.

Fourth aspect, it is provided that the overall Buffer Pool management method of a kind of cluster multiple buffer pool system, wherein This cluster multiple buffer pool system includes Centroid, overall situation Buffer Pool and calculates node, this overall situation Buffer Pool Including the buffer joint independent of this Centroid, the method includes: this calculating node is to this Centroid Sending page request, this page request carries node serial number and this calculating node application visit of this calculating node The page number of the memory pages asked；This calculating node receives this memory pages of this Centroid transmission and exists Storage positional information in this overall situation Buffer Pool, wherein, this memory pages depositing in this overall situation Buffer Pool Storage positional information is by this Centroid node serial number according to this calculating node and the page of this memory pages Numbering and obtain from this overall situation Buffer Pool, this overall situation Buffer Pool is for storing the complete of this cluster multiple buffer pool system Office's shared buffer data；This calculating node is according to this memory pages storage position in this overall situation Buffer Pool Information, conducts interviews to this memory pages.

In conjunction with fourth aspect, in the implementation that the first is possible, this overall situation Buffer Pool includes at least 2 buffer joints, store the complete of this overall situation Buffer Pool in all buffer joints in this overall situation Buffer Pool At least 2 backups of office's shared buffer data, this memory pages storage position in this overall situation Buffer Pool Information includes the storage position of this memory pages at least two buffer joint in this overall situation Buffer Pool, should Calculate node according to this memory pages storage positional information in this overall situation Buffer Pool, to this memory pages Conduct interviews and be implemented as: when this memory pages is read out accessing by this calculating node, this calculating Node select this memory pages at least two buffer joint in this overall situation Buffer Pool storage position it A pair this memory pages is read out accessing；Or visit when this calculating node carries out write to this memory pages When asking, this calculating node selects this memory pages at least two buffer joint in this overall situation Buffer Pool Storage position carries out write-access to this memory pages.

5th aspect, it is provided that the Centroid of a kind of cluster multiple buffer pool system, wherein this cluster delays more Rushing cell system include this Centroid, overall situation Buffer Pool and calculate node, this overall situation Buffer Pool includes independence In the buffer joint of this Centroid, this Centroid includes: receive unit, is used for receiving this calculating joint The page request that point sends, this page request carries node serial number and this calculating node Shen of this calculating node The page number of the memory pages that please access；Acquiring unit, for the node serial number according to this calculating node With the page number of this memory pages, obtain this memory pages storage position letter in this overall situation Buffer Pool Breath, wherein, this overall situation Buffer Pool is for storing the globally shared buffered data of this cluster multiple buffer pool system； Transmitting element, for sending this memory pages storage position in this overall situation Buffer Pool to this calculating node Information, wherein, this memory pages storage positional information in this overall situation Buffer Pool is used for this calculating node This memory pages is conducted interviews.

In conjunction with the 5th aspect, in the implementation that the first is possible, it is implemented as: this Centroid Also include administrative unit, for managing the buffer joint of this overall situation Buffer Pool.

In conjunction with the first possible implementation of the 5th aspect, in the implementation that the second is possible, This administrative unit is implemented as: the registration request being received the first buffer joint by this reception unit is believed Breath, this registration information carries node serial number and the memory source information of this first buffer joint, wherein, This first buffer joint is the buffer joint outside this overall situation Buffer Pool；Internal memory according to this first buffer joint The memory source of this first buffer joint is registered in this overall situation Buffer Pool by resource information.

In conjunction with the first possible implementation of the 5th aspect, in the implementation that the third is possible, This administrative unit is implemented as: receive, by this reception unit, the de-registration request that the second buffer joint sends Information, this de-registration request information carries the node serial number of this second buffer joint, wherein, this second buffering Node is the buffer joint in this overall situation Buffer Pool；According to this de-registration request information by this second buffer joint Data Migration in the buffer joint beyond this second buffer joint in this overall situation Buffer Pool；By this second The memory source information of buffer joint is deleted from this overall situation Buffer Pool.

In conjunction with the first possible implementation of the 5th aspect, in the 4th kind of possible implementation, This administrative unit is implemented as: if this reception unit receives leading to about the second buffer joint fault Know, then according to this notice by the Data Migration of this second buffer joint to this overall situation Buffer Pool, this second delays Rush in the buffer joint beyond node, and the memory source information of this second buffer joint is delayed from this overall situation Rushing in pond and delete, wherein, this second buffer joint is the buffer joint in this overall situation Buffer Pool.

In conjunction with the first possible implementation of the 5th aspect, in the 5th kind of possible implementation, This administrative unit is implemented as: receive, by this reception unit, the replacement request that the second buffer joint sends Information, this replacement request information carries the memory source information of the 3rd buffer joint, and wherein, this second delays Rushing node is the buffer joint in this overall situation Buffer Pool, and the 3rd buffer joint is outside this overall situation Buffer Pool Buffer joint；According to this replacement request information by the Data Migration of this second buffer joint to the 3rd buffering In node, and delete the log-on message of this second buffer joint, and according to the internal memory of the 3rd buffer joint The memory source of the 3rd buffer joint is registered to this overall situation Buffer Pool by resource information.

In conjunction with the first possible implementation of the 5th aspect or the 5th aspect to the 5th of the 5th aspect Plant any one possible implementation in possible implementation, in the 6th kind of possible implementation, It is implemented as: this overall situation Buffer Pool includes at least 2 buffer joints, the institute in this overall situation Buffer Pool There are in buffer joint at least 2 backups of the globally shared buffered data that store this overall situation Buffer Pool, should Memory pages storage positional information in this overall situation Buffer Pool includes that this memory pages buffers in this overall situation The storage position of at least two buffer joint in pond.

6th aspect, it is provided that the calculating node of a kind of cluster multiple buffer pool system, wherein this cluster delays more Rush cell system include Centroid, the overall situation Buffer Pool and calculate node, this overall situation Buffer Pool include independent of The buffer joint of this Centroid, this calculating node includes: transmitting element, for sending out to this Centroid Sending page request, this page request carries node serial number and this calculating node application access of this calculating node The page number of memory pages；Receive unit, for receiving this memory pages that this Centroid sends Storage positional information in this overall situation Buffer Pool, wherein, this memory pages is in this overall situation Buffer Pool Storage positional information is by this Centroid node serial number according to this calculating node and the page of this memory pages Face numbering obtains from this overall situation Buffer Pool, and this overall situation Buffer Pool is for storing this cluster multiple buffer pool system Globally shared buffered data；Access unit, for according to this memory pages depositing in this overall situation Buffer Pool Storage positional information, conducts interviews to this memory pages.

In conjunction with the 6th aspect, in the implementation that the first is possible, this overall situation Buffer Pool include to Few 2 buffer joints, store this overall situation Buffer Pool in all buffer joints in this overall situation Buffer Pool At least 2 backups of globally shared buffered data, this memory pages storage position in this overall situation Buffer Pool Confidence breath includes the storage position of this memory pages at least two buffer joint in this overall situation Buffer Pool, This access unit is implemented as: when being read out this memory pages accessing, select this memory pages A pair this memory pages of the storage position of at least two buffer joint in this overall situation Buffer Pool is carried out Read access；Or when this memory pages is carried out write-access, select this memory pages in this overall situation The storage position of at least two buffer joint in Buffer Pool carries out write-access to this memory pages.

Based on above technical scheme, cluster many overall situations Buffer Pool system of the embodiment of the present invention, Centroid, Calculating node and management method, Centroid and overall situation Buffer Pool separate, Centroid is according to calculating The request of node is managed distribution to the resource of overall situation Buffer Pool, reduces overall situation Buffer Pool and centromere The degree of coupling between point, so that overall situation Buffer Pool possesses the possibility of flexible expansion.

Accompanying drawing explanation

In order to be illustrated more clearly that the technical scheme of the embodiment of the present invention, below will be to embodiment or existing skill In art description, the required accompanying drawing used is briefly described, it should be apparent that, the accompanying drawing in describing below It is only some embodiments of the present invention, for those of ordinary skill in the art, is not paying creation Property work on the premise of, it is also possible to obtain other accompanying drawing according to these accompanying drawings.

Fig. 1 is embodiment of the present invention overall situation Buffer Pool management method flow chart.

Fig. 2 is the interaction diagrams of embodiment of the present invention overall situation Buffer Pool management.

Fig. 3 is the interaction diagrams of embodiment of the present invention registration buffer node.

Fig. 4 is the interaction diagrams that the embodiment of the present invention nullifies buffer joint.

Fig. 5 is the interaction diagrams that the embodiment of the present invention deletes fault buffer joint.

Fig. 6 is the interaction diagrams that the embodiment of the present invention changes buffer joint.

Fig. 7 is another management method flow chart of embodiment of the present invention overall situation Buffer Pool.

Fig. 8 is embodiment of the present invention overall situation Buffer Pool another management method flow chart.

Fig. 9 is the structural representation of embodiment of the present invention cluster multiple buffer pool system.

Figure 10 is the structural representation of embodiment of the present invention Centroid.

Figure 11 is the structural representation that the embodiment of the present invention calculates node.

Figure 12 is another structural representation of embodiment of the present invention Centroid.

Figure 13 is another structural representation that the embodiment of the present invention calculates node.

Detailed description of the invention

Below in conjunction with the accompanying drawing in the embodiment of the present invention, the technical scheme in the embodiment of the present invention is carried out Clearly and completely describe, it is clear that described embodiment be a part of embodiment of the present invention rather than Whole embodiments.Based on the embodiment in the present invention, those of ordinary skill in the art are not making wound The every other embodiment obtained under the property made work premise, broadly falls into the scope of protection of the invention.

To facilitate understanding of the present embodiment of the invention, first introduce at this during embodiment of the present invention describes and can introduce Several key elements.

Cluster, the parallel or distributed system being made up of the computer interconnected, can as individually, Unified calculating resource uses.Group system (Clusting System) refers to two-shipper or the multimachine of equally loaded System.

Centroid, deposits the node of the public informations such as all log-on messages, and each calculating node all relies on Centroid completes to access.

Calculate node, perform the node of customer service logic.

Overall situation Buffer Pool, is made up of buffer joint, is used for depositing globally shared buffered data.

Fig. 1 is embodiment of the present invention overall situation Buffer Pool management method flow chart.Wherein, cluster multiple buffer pool System includes Centroid, overall situation Buffer Pool and calculates node, and this overall situation Buffer Pool includes independent of in this The buffer joint of heart node.The method includes:

101, this calculating node sends page request to this Centroid.

Wherein, this page request carries node serial number and this calculating node application access of this calculating node The page number of memory pages.

102, this Centroid node serial number according to this calculating node and the page number of this memory pages, This memory pages storage positional information in this overall situation Buffer Pool is provided to this calculating node.

Wherein, this overall situation Buffer Pool is for storing the globally shared buffer number of this cluster multiple buffer pool system According to.

103, this calculating node according to this memory pages storage positional information in this overall situation Buffer Pool, This memory pages is conducted interviews.

In the embodiment of the present invention, the Centroid of cluster multiple buffer pool system and overall situation Buffer Pool are the most only Vertical, Centroid is managed distribution according to the request calculating node to the resource of overall situation Buffer Pool, reduces The degree of coupling between overall Buffer Pool and Centroid, so that overall situation Buffer Pool possesses flexible expansion Possibility.

It addition, the Centroid of cluster multiple buffer pool system is by calculating the application page request message of node For calculate node locating access memory pages the overall situation Buffer Pool in position in case calculate node can The memory source of overall situation Buffer Pool is accessed flexibly, the internal memory money of the Buffer Pool of the management overall situation effectively Source, improves the utilization ratio of overall situation Buffer Pool.

Alternatively, the method also includes: this Centroid manages the buffer joint of this overall situation Buffer Pool.Collection The buffer joint of overall situation Buffer Pool is managed by group's multiple buffer pool system by Centroid, it is possible to flexibly Ground adjusts the buffer joint of overall situation Buffer Pool.It addition, cluster multiple buffer pool system passes through Centroid to entirely Office's Buffer Pool is managed, and also can reduce in the overall situation Buffer Pool additions and deletions of buffer joint to a certain extent to being The adverse effect that regiment commander comes.

Alternatively, as an embodiment, this Centroid manages the buffer joint tool of this overall situation Buffer Pool Body is embodied as: this Centroid receives the registration information of the first buffer joint, this registration information Carrying node serial number and the memory source information of this first buffer joint, wherein, this first buffer joint is Buffer joint outside this overall situation Buffer Pool；This Centroid is believed according to the memory source of this first buffer joint The memory source of this first buffer joint is registered in this overall situation Buffer Pool by breath.In the embodiment of the present invention, Cluster multiple buffer pool system can complete the extension to overall situation Buffer Pool according to the registration request of buffer joint, real Show the flexible expansion of cluster overall situation Buffer Pool.Specifically, this memory source information includes this first buffering The memory address of node, internal memory length and remote direct memory access (Remote Direct Memory Access, RDMA) access key KEY, in order to the Centroid internal memory to newly-increased buffer joint Resource more fully manages.

Alternatively, as another embodiment, this Centroid manages the buffer joint of this overall situation Buffer Pool Being implemented as: this Centroid receives the de-registration request information that the second buffer joint sends, this cancellation please Asking information to carry the node serial number of this second buffer joint, wherein, this second buffer joint is that this overall situation is delayed Rush the buffer joint in pond；This Centroid according to this de-registration request information by the number of this second buffer joint According in the buffer joint moved in this overall situation Buffer Pool beyond this second buffer joint；This Centroid will The memory source information of this second buffer joint is deleted from this overall situation Buffer Pool.In the embodiment of the present invention, Cluster multiple buffer pool system can be nullified slow according to the de-registration request information of buffer joint in overall situation Buffer Pool Rush node, it is possible to the buffer joint of overall situation Buffer Pool is deleted flexibly.

Alternatively, as further embodiment, this Centroid manages the buffer joint of this overall situation Buffer Pool It is implemented as: if this Centroid receives the notice about the second buffer joint fault, then in this Heart node according to this notice by the Data Migration of this second buffer joint to this overall situation Buffer Pool this second In buffer joint beyond buffer joint, and by the memory source information of this second buffer joint from this overall situation Deleting in Buffer Pool, wherein, this second buffer joint is the buffer joint in this overall situation Buffer Pool.This In bright embodiment, cluster multiple buffer pool system can buffer in the overall situation according to the failure notification about buffer joint Pond is nullified buffer joint, it is possible to the fault buffer joint of overall situation Buffer Pool is deleted flexibly.

Alternatively, as further embodiment, this Centroid manages the buffer joint of this overall situation Buffer Pool Being implemented as: this Centroid receives the replacement request information that the second buffer joint sends, this replacing please Asking information to carry the memory source information of the 3rd buffer joint, wherein, this second buffer joint is this overall situation Buffer joint in Buffer Pool, the 3rd buffer joint is the buffer joint outside this overall situation Buffer Pool；In this The Data Migration of this second buffer joint to the 3rd buffering is saved by heart node according to this replacement request information In point, and delete the log-on message of this second buffer joint, and provide according to the internal memory of the 3rd buffer joint The memory source of the 3rd buffer joint is registered to this overall situation Buffer Pool by source information.The embodiment of the present invention In, cluster multiple buffer pool system can be according to the replacement request information realization buffer joint about buffer joint Change, it is possible to the buffer joint of overall situation Buffer Pool is replaced flexibly.

Alternatively, as further embodiment, this Centroid manages the buffer joint of this overall situation Buffer Pool It is implemented as: this overall situation Buffer Pool includes at least 2 buffer joints, the institute in this overall situation Buffer Pool There are in buffer joint at least 2 backups of the globally shared buffered data that store this overall situation Buffer Pool.This In inventive embodiments, overall situation Buffer Pool includes multiple buffer joint, and stores the multiple of overall situation Buffer Pool Data backup, when calculating node visit and being more frequent, can improve the efficiency calculating node visit, Jin Erti The systematic function of high cluster multiple buffer pool system.

Further, this memory pages storage positional information in this overall situation Buffer Pool includes this page The storage position of face at least two buffer joint in this overall situation Buffer Pool, this calculating node is interior according to this Deposit page storage positional information in this overall situation Buffer Pool this memory pages is conducted interviews to implement For: when this memory pages is read out accessing by this calculating node, this calculating node selects this page A pair this memory pages of the storage position of face at least two buffer joint in this overall situation Buffer Pool enters Row read access；Or when this calculating node carries out write-access to this memory pages, this calculating node The storage position selecting this memory pages at least two buffer joint in this overall situation Buffer Pool is interior to this Deposit the page and carry out write-access.

Alternatively, this application page request message also carries the page byte number of this memory pages, this page The page number that byte number combines this memory pages for Centroid 901 determines this storage positional information.

Below, will in conjunction with specific embodiments, the system of the embodiment of the present invention etc. is further described.

Below in conjunction with specific embodiments, the method for the embodiment of the present invention is described.

Fig. 2 is the interaction diagrams of embodiment of the present invention overall situation Buffer Pool management.Centroid in Fig. 2 For the Centroid of embodiment of the present invention cluster multiple buffer pool system, overall situation Buffer Pool is the embodiment of the present invention The overall Buffer Pool of cluster multiple buffer pool system, calculating node is embodiment of the present invention cluster multiple buffer pool system The calculating node of system, buffer joint B, D are the buffer joint in overall situation Buffer Pool.

201, calculate node and send application page request message.

Calculate node and send application page request message to Centroid.Wherein, this application page request letter Breath can carry the page number of the memory pages of request and the node serial number of this calculating node.Certainly, should Application page request message also can carry the page length of this memory pages, such as, this memory pages Total bytes, etc..

202, Centroid obtains the memory pages storage position at overall situation Buffer Pool.

The page number of the memory pages that Centroid can carry according to application page request message, obtaining should Memory pages is in the storage positional information of overall situation Buffer Pool.

Generally, cluster there may be the backup of many sets of data, deposit in different buffer joint, Therefore, this storage positional information potentially includes the storage positional information of multiple buffer joint.

Such as, in the embodiment of the present invention, this storage positional information can include that this memory pages is at buffer joint The storage positional information of B and buffer joint D.

203, Centroid sends request confirmation to calculating node.

Centroid get this memory pages the overall situation Buffer Pool storage positional information after, can be to calculating Node sends the request confirmation of application page request message, and wherein, this request confirmation can be carried This memory pages is in the storage positional information of overall situation Buffer Pool.

204, calculate node and conduct interviews in the storage positional information of overall situation Buffer Pool according to memory pages.

The access calculating node can include that 2 classes, a class are read operations, and a class is write operation.Work as meter When the access of operator node is read operation, calculate node can from this memory pages the overall situation Buffer Pool storage The storage positional information of the multiple buffer joints comprised in positional information select one be read out accessing. When the access calculating node is write operation, calculating node needs memory pages at overall situation Buffer Pool Each of the storage positional information of the multiple buffer joints comprised in storage positional information is write respectively Enter operation.

In the embodiment of the present invention, Centroid is by calculating the application page request message of node for calculating joint The memory pages that point location accesses position in overall situation Buffer Pool is to calculate node and can delaying the overall situation The memory source rushing pond accesses flexibly, the memory source of the Buffer Pool of the management overall situation effectively, improves The utilization ratio of overall situation Buffer Pool.It addition, Centroid and overall situation Buffer Pool are separate, it is possible to reduce The degree of coupling between Centroid and overall situation Buffer Pool, thus decrease the generation that overall situation buffer pool extension needs Valency.

Fig. 3 is the interaction diagrams of embodiment of the present invention registration buffer node.Centroid in Fig. 3 is The Centroid of embodiment of the present invention cluster multiple buffer pool system, overall situation Buffer Pool is embodiment of the present invention collection The overall Buffer Pool of group's multiple buffer pool system, buffer joint A is the buffer joint outside overall situation Buffer Pool.

301, buffer joint A send registration information.

The registration information of buffer joint A can include node serial number and the memory source of buffer joint A Information.Wherein, memory source information can include the memory address of buffer joint A, internal memory length and RDMA Key etc..Specifically, internal memory length can be the total bytes of internal memory.

302, buffer joint A is registered to overall situation Buffer Pool by Centroid.

Centroid, can be according to buffer joint after receiving the registration information that buffer joint A sends The memory source information of A, is registered to the memory source of buffer joint A in overall situation Buffer Pool.In Tong Shi Heart node also can carry out resource division to the memory source of registration and (such as, memory source be carried out paging behaviour Make), for the calculating node application resource of Centroid place cluster.

303, Centroid is by the part Data Migration of overall situation Buffer Pool to buffer joint A.

After buffer joint A adds overall situation Buffer Pool, Centroid can start data in due course Migrate, by the part Data Migration of other node of overall situation Buffer Pool to buffer joint A.Such as, center Node can start Data Migration in the case of light load.

In the embodiment of the present invention, Centroid completes overall situation Buffer Pool according to the registration request of buffer joint Extension, it is achieved that the flexible expansion of cluster overall situation Buffer Pool.

Fig. 4 is the interaction diagrams that the embodiment of the present invention nullifies buffer joint.Centroid in Fig. 4 is The Centroid of embodiment of the present invention cluster multiple buffer pool system, overall situation Buffer Pool is embodiment of the present invention collection The overall Buffer Pool of group's multiple buffer pool system, buffer joint B is the buffer joint in overall situation Buffer Pool.

401, buffer joint B send de-registration request information.

Wherein, this de-registration request information can carry the node serial number of buffer joint B, this de-registration request information For asking Centroid to nullify buffer joint B from overall situation Buffer Pool.

402, Centroid is by other buffer joint of the Data Migration of buffer joint B to overall situation Buffer Pool In.

Centroid, after receiving de-registration request information, can start Data Migration to buffer joint B, will In buffer joint B store Data Migration to the overall situation Buffer Pool other buffer joint in.

403, Centroid nullifies buffer joint B from overall situation Buffer Pool.

Centroid, after migrating, can delete buffer joint B memory source in overall situation Buffer Pool Information, thus nullify buffer joint B in overall situation Buffer Pool.

In the embodiment of the present invention, Centroid is delayed in the overall situation by the de-registration request information according to buffer joint Rush cancellation buffer joint in pond, it is possible to the buffer joint of overall situation Buffer Pool is deleted flexibly.

Fig. 5 is the interaction diagrams that the embodiment of the present invention deletes fault buffer joint.Centromere in Fig. 5 Point is the Centroid of embodiment of the present invention cluster multiple buffer pool system, and overall situation Buffer Pool is that the present invention implements The overall Buffer Pool of example cluster multiple buffer pool system, buffer joint B is the buffering joint in overall situation Buffer Pool Point.

501, Centroid receives the failure notification about buffer joint B.

Centroid receives the failure notification about buffer joint B of Centroid place collection pocket transmission. Wherein, this failure notification can carry the node serial number of buffer joint B.Certainly, this failure notification also can be taken The type etc. of tape jam, this is not restricted for the embodiment of the present invention.

502, Centroid is by other buffer joint of the Data Migration of buffer joint B to overall situation Buffer Pool In.

Centroid upon receipt of the failure notification, can start Data Migration to buffer joint B, will buffering In node B store Data Migration to the overall situation Buffer Pool other buffer joint in.

503, Centroid nullifies buffer joint B from overall situation Buffer Pool.

In the embodiment of the present invention, Centroid is by delaying in the overall situation according to the failure notification about buffer joint Rush cancellation buffer joint in pond, it is possible to the buffer joint of overall situation Buffer Pool is deleted flexibly.

Fig. 6 is the interaction diagrams that the embodiment of the present invention changes buffer joint.Centroid in Fig. 6 is The Centroid of embodiment of the present invention cluster multiple buffer pool system, overall situation Buffer Pool is embodiment of the present invention collection The overall Buffer Pool of group's multiple buffer pool system, buffer joint B is the buffer joint in overall situation Buffer Pool, slow Rushing node C is the buffer joint outside overall situation Buffer Pool.

601, buffer joint B send replacement request information.

Buffer joint B sends replacement request information to Centroid, and buffer joint B is replaced by by request Buffer joint C.Wherein, this replacement request information can carry the node serial number of buffer joint B, buffering joint The node serial number of some C and the memory source information of buffer joint C.

602, Centroid is registration buffer node C in overall situation Buffer Pool.

The memory source of buffer joint C is registered by Centroid.

603, Centroid is by the Data Migration of buffer joint B to buffer joint C.

Centroid, after buffer joint C registers, can start Data Migration, by buffer joint B Data Migration in buffer joint C.

604, Centroid nullifies buffer joint B from overall situation Buffer Pool.

After Data Migration, Centroid can delete the log-on message of buffer joint B, thus entirely Office's Buffer Pool nullifies buffer joint B.

In the embodiment of the present invention, Centroid is by according to the replacement request information realization about buffer joint The replacing of buffer joint, it is possible to the buffer joint of overall situation Buffer Pool is replaced flexibly.

Fig. 7 is the overall Buffer Pool management method flow chart of embodiment of the present invention cluster multiple buffer pool system. The method of Fig. 7 is performed by Centroid.Wherein, cluster multiple buffer pool system includes Centroid, the overall situation Buffer Pool and calculating node, this overall situation Buffer Pool includes the buffer joint independent of this Centroid.The party Method includes:

701, this Centroid receives the page request that this calculating node sends.

702, this Centroid node serial number according to this calculating node and the page number of this memory pages, Obtain this memory pages storage positional information in this overall situation Buffer Pool.

703, this Centroid sends this memory pages depositing in this overall situation Buffer Pool to this calculating node Storage positional information.

Wherein, this memory pages storage positional information in this overall situation Buffer Pool is used for this calculating node pair This memory pages conducts interviews.

In the embodiment of the present invention, Centroid and overall situation Buffer Pool are separate, and Centroid is according to calculating The request of node is managed distribution to the resource of overall situation Buffer Pool, reduces overall situation Buffer Pool and centromere The degree of coupling between point, so that overall situation Buffer Pool possesses the possibility of flexible expansion.

It addition, Centroid accesses for calculating node locating by calculating the application page request message of node Memory pages the overall situation Buffer Pool in position in case calculate node can to the overall situation Buffer Pool internal memory Resource accesses flexibly, the memory source of the Buffer Pool of the management overall situation effectively, improves overall situation Buffer Pool Utilization ratio.

Alternatively, the method also includes: this Centroid manages the buffer joint of this overall situation Buffer Pool.In The buffer joint of overall situation Buffer Pool is managed by heart node, it is possible to adjust the slow of overall situation Buffer Pool neatly Rush node.It addition, overall situation Buffer Pool is managed by Centroid, the overall situation also can be reduced to a certain extent The adverse effect that in Buffer Pool, system is brought by the additions and deletions of buffer joint.

Alternatively, as an embodiment, this Centroid manages the buffer joint tool of this overall situation Buffer Pool Body is embodied as: this Centroid receives the registration information of the first buffer joint, this registration information Carrying node serial number and the memory source information of this first buffer joint, wherein, this first buffer joint is Buffer joint outside this overall situation Buffer Pool；This Centroid is believed according to the memory source of this first buffer joint The memory source of this first buffer joint is registered in this overall situation Buffer Pool by breath.In the embodiment of the present invention, Centroid can complete the extension to overall situation Buffer Pool according to the registration request of buffer joint, it is achieved that cluster The flexible expansion of overall situation Buffer Pool.Specifically, this memory source information includes the interior of this first buffer joint Deposit address, internal memory length and remote direct memory and access RDMA access key KEY, in order to center The memory source of newly-increased buffer joint is more fully managed by node.

Alternatively, as another embodiment, this Centroid manages the buffer joint of this overall situation Buffer Pool Being implemented as: this Centroid receives the de-registration request information that the second buffer joint sends, this cancellation please Asking information to carry the node serial number of this second buffer joint, wherein, this second buffer joint is that this overall situation is delayed Rush the buffer joint in pond；This Centroid according to this de-registration request information by the number of this second buffer joint According in the buffer joint moved in this overall situation Buffer Pool beyond this second buffer joint；This Centroid will The memory source information of this second buffer joint is deleted from this overall situation Buffer Pool.In the embodiment of the present invention, Centroid can nullify buffer joint, energy according to the de-registration request information of buffer joint in overall situation Buffer Pool Enough buffer joints to overall situation Buffer Pool are deleted flexibly.

Alternatively, as further embodiment, this Centroid manages the buffer joint of this overall situation Buffer Pool It is implemented as: if this Centroid receives the notice about the second buffer joint fault, then in this Heart node according to this notice by the Data Migration of this second buffer joint to this overall situation Buffer Pool this second In buffer joint beyond buffer joint, and by the memory source information of this second buffer joint from this overall situation Deleting in Buffer Pool, wherein, this second buffer joint is the buffer joint in this overall situation Buffer Pool.This In bright embodiment, Centroid can be nullified in overall situation Buffer Pool according to the failure notification about buffer joint Buffer joint, it is possible to the fault buffer joint of overall situation Buffer Pool is deleted flexibly.

Alternatively, as further embodiment, this Centroid manages the buffer joint of this overall situation Buffer Pool Being implemented as: this Centroid receives the replacement request information that the second buffer joint sends, this replacing please Asking information to carry the memory source information of the 3rd buffer joint, wherein, this second buffer joint is this overall situation Buffer joint in Buffer Pool, the 3rd buffer joint is the buffer joint outside this overall situation Buffer Pool；In this The Data Migration of this second buffer joint to the 3rd buffering is saved by heart node according to this replacement request information In point, and delete the log-on message of this second buffer joint, and provide according to the internal memory of the 3rd buffer joint The memory source of the 3rd buffer joint is registered to this overall situation Buffer Pool by source information.The embodiment of the present invention In, Centroid can be according to the replacing of the replacement request information realization buffer joint about buffer joint, energy Enough buffer joints to overall situation Buffer Pool are replaced flexibly.

Alternatively, this overall situation Buffer Pool includes at least 2 buffer joints, the institute in this overall situation Buffer Pool There are in buffer joint at least 2 backups of the globally shared buffered data that store this overall situation Buffer Pool, should Memory pages storage positional information in this overall situation Buffer Pool includes that this memory pages buffers in this overall situation The storage position of at least two buffer joint in pond.In the embodiment of the present invention, overall situation Buffer Pool includes Multiple buffer joints, and store multiple data backups of overall situation Buffer Pool, calculating, node visit is more frequent Time, the efficiency calculating node visit can be improved, and then improve the systematic function of cluster multiple buffer pool system.

Alternatively, this application page request message also carries the page byte number of this memory pages, this page The page number that byte number combines this memory pages for Centroid determines this storage positional information.

It addition, implementing of the method for the embodiment of the present invention, refer to the enforcement shown in Fig. 2 to Fig. 6 The method that example Centroid performs, the embodiment of the present invention does not repeats them here.

Fig. 8 is the overall Buffer Pool management method flow chart of embodiment of the present invention cluster multiple buffer pool system. The method of Fig. 8 is performed by calculating node.Wherein, cluster multiple buffer pool system includes Centroid, the overall situation Buffer Pool and calculating node, this overall situation Buffer Pool includes the buffer joint independent of this Centroid.The party Method includes:

801, this calculating node sends page request to this Centroid.

802, this calculating node receives this memory pages of this Centroid transmission in this overall situation Buffer Pool Storage positional information.

Wherein, this memory pages storage positional information in this overall situation Buffer Pool by this Centroid according to The node serial number of this calculating node and the page number of this memory pages obtain from this overall situation Buffer Pool, and this is complete Office's Buffer Pool is for storing the globally shared buffered data of this cluster multiple buffer pool system.

803, this calculating node according to this memory pages storage positional information in this overall situation Buffer Pool, This memory pages is conducted interviews.

The embodiment of the present invention, Centroid and overall situation Buffer Pool are separate, calculate node by center Node sends page request and obtains this memory pages position at overall situation Buffer Pool so that calculate node request The memory source of Centroid it is not necessarily limited to, so that the overall situation is slow during overall situation Buffer Pool memory source Rush pond and possess the possibility of flexible expansion.

Alternatively, this overall situation Buffer Pool includes at least 2 buffer joints, the institute in this overall situation Buffer Pool There are in buffer joint at least 2 backups of the globally shared buffered data that store this overall situation Buffer Pool, should Memory pages storage positional information in this overall situation Buffer Pool includes that this memory pages buffers in this overall situation The storage position of at least two buffer joint in pond.

Further, this calculating node according to this memory pages in this overall situation Buffer Pool storage position letter Breath, conducts interviews to this memory pages and is implemented as: when this memory pages is read by this calculating node When taking access, this calculating node selects this memory pages at least two buffering joint in this overall situation Buffer Pool A pair this memory pages of the storage position of point is read out accessing；Or when this calculating node is to this internal memory When the page carries out write-access, this calculating node select this memory pages in this overall situation Buffer Pool at least The storage position of two buffer joints carries out write-access to this memory pages.

It addition, implementing of the method for the embodiment of the present invention, refer to the embodiment center shown in Fig. 2 The method that node performs, the embodiment of the present invention does not repeats them here.

Fig. 9 is the structural representation of embodiment of the present invention cluster multiple buffer pool system 900.How overall cluster is Buffer Pool system 900 can include Centroid 901, overall situation Buffer Pool 902 and calculate node 903, its In,

Overall situation Buffer Pool 902, can be used for the globally shared buffer number of storage cluster multiple buffer pool system 900 According to.In overall situation Buffer Pool 902, it may include independent of the buffer joint of Centroid 901.

Certainly, if Centroid there is also the device being similar to buffer joint, overall situation Buffer Pool 902 May also comprise this device, this is not restricted for the embodiment of the present invention.

Calculate node 903, for sending page request to Centroid 901.

Wherein, this page request is carried to calculate the node serial number of node 903 and calculate node 903 application and is visited The page number of the memory pages asked.

Centroid 901, can compile according to the page of the node serial number and this memory pages that calculate node 903 Number, provide this memory pages storage positional information in overall situation Buffer Pool 902 to calculating node 903.

Specifically, Centroid 901 can be according to node serial number and this memory pages calculating node 903 Page number, obtains this memory pages storage positional information in overall situation Buffer Pool 902, and this is interior Deposit page storage positional information in overall situation Buffer Pool 902 to be sent to calculate node 903.

Calculate node 903 and be additionally operable to this memory pages according to Centroid 901 offer at overall situation Buffer Pool Storage positional information in 902, conducts interviews to this memory pages.

Alternatively, Centroid 901 is additionally operable to manage the buffer joint of overall situation Buffer Pool 902.Cluster is many Overall situation Buffer Pool is managed by Buffer Pool system 900 by Centroid, it is possible to adjust the overall situation neatly The buffer joint of Buffer Pool.It addition, the overall situation is buffered by cluster multiple buffer pool system 900 by Centroid Pond is managed, and also can reduce the additions and deletions of buffer joint in overall situation Buffer Pool to a certain extent and bring system Adverse effect.

Alternatively, as an embodiment, at the buffer joint for managing overall situation Buffer Pool 902, in Heart node 901 specifically for: receive the registration information of the first buffer joint, this registration information Carrying node serial number and the memory source information of this first buffer joint, wherein, this first buffer joint is Buffer joint outside overall situation Buffer Pool 902；Memory source information according to this first buffer joint by this The memory source of one buffer joint is registered in overall situation Buffer Pool 902.In the embodiment of the present invention, cluster is many Buffer Pool system 900 can complete the extension to overall situation Buffer Pool according to the registration request of buffer joint, it is achieved The flexible expansion of cluster overall situation Buffer Pool.Specifically, this memory source information includes this first buffering joint The memory address of point, internal memory length and remote direct memory access RDMA access key KEY, with Just the memory source of newly-increased buffer joint is more fully managed by Centroid.

Alternatively, as another embodiment, at the buffer joint for managing overall situation Buffer Pool 902, Centroid 901 specifically for: receive second buffer joint send de-registration request information, this cancellation please Asking information to carry the node serial number of this second buffer joint, wherein, this second buffer joint is overall situation buffering Buffer joint in pond 902；According to this de-registration request information, the Data Migration of this second buffer joint is arrived In overall situation Buffer Pool 902 in buffer joint beyond this second buffer joint；By this second buffer joint Memory source information is deleted from overall situation Buffer Pool 902.In the embodiment of the present invention, cluster multiple buffer pool system System 900 can nullify buffer joint according to the de-registration request information of buffer joint in overall situation Buffer Pool, it is possible to The buffer joint of overall situation Buffer Pool is deleted flexibly.

Alternatively, as further embodiment, at the buffer joint for managing overall situation Buffer Pool 902, Centroid 901 specifically for: if receiving the notice about the second buffer joint fault, then according to should Notify in the buffer joint beyond by the Data Migration of this second buffer joint to overall situation Buffer Pool 902, and The memory source information of this second buffer joint being deleted from overall situation Buffer Pool 902, wherein, this is second years old Buffer joint is the buffer joint in overall situation Buffer Pool 902.In the embodiment of the present invention, cluster multiple buffer pool System 900 can nullify buffer joint, energy according to the failure notification about buffer joint in overall situation Buffer Pool Enough fault buffer joints to overall situation Buffer Pool are deleted flexibly.

Alternatively, as further embodiment, at the buffer joint for managing overall situation Buffer Pool 902, Centroid 901 specifically for: receive second buffer joint send replacement request information, this replacing please Asking information to carry the memory source information of the 3rd buffer joint, wherein, this second buffer joint is that the overall situation is slow Rushing the buffer joint in pond 902, the 3rd buffer joint is the buffer joint outside overall situation Buffer Pool 902； According to this replacement request information by the Data Migration of this second buffer joint to the 3rd buffer joint, and Delete the log-on message of this second buffer joint, and the memory source information according to the 3rd buffer joint will The memory source of the 3rd buffer joint is registered to overall situation Buffer Pool 902.In the embodiment of the present invention, cluster Multiple buffer pool system 900 can according to the replacement request information realization buffer joint about buffer joint more Change, it is possible to the buffer joint of overall situation Buffer Pool is replaced flexibly.

Alternatively, overall situation Buffer Pool 902 includes at least 2 buffer joints, in overall situation Buffer Pool 902 All buffer joints in store at least 2 of globally shared buffered data of overall situation Buffer Pool 902 standby Part.In the embodiment of the present invention, overall situation Buffer Pool includes multiple buffer joint, and stores overall situation Buffer Pool Multiple data backups, calculate node visit more frequent time, can improve calculate node visit efficiency, And then improve the systematic function of cluster multiple buffer pool system 900.

Further, this memory pages storage positional information in overall situation Buffer Pool 902 includes this internal memory The storage position of page at least two buffer joint in overall situation Buffer Pool 902.Calculate node 903 According to Centroid 901 provide this memory pages the overall situation Buffer Pool 902 in storage positional information to this Memory pages conducts interviews and includes: when calculating node 903 and being read out this memory pages accessing, meter Operator node 903 selects the storage of this memory pages at least two buffer joint in overall situation Buffer Pool 902 A pair of position this memory pages is read out accessing；Or when calculating node 903, this memory pages is entered During row write-access, calculate node 903 and select this memory pages at least two in overall situation Buffer Pool 902 The storage position of individual buffer joint carries out write-access to this memory pages.

In the embodiment of the present invention, cluster multiple buffer pool system 900 can include one or more calculating node, May also include one or more Centroid.When cluster multiple buffer pool system 900 has multiple Centroid Time, Centroid can be divided into main Centroid and from Centroid, and wherein, main Centroid has this All functions of bright embodiment Centroid 901, can have the part of Centroid 901 from Centroid Function, such as, according to the memory source of calculating node request allocation manager overall situation Buffer Pool, etc..

It addition, the method that cluster multiple buffer pool system 900 can also carry out Fig. 1, it implements and refers to Fig. 1 does not repeats them here to embodiment illustrated in fig. 6, the embodiment of the present invention.

Figure 10 is the structural representation of the Centroid 1000 of embodiment of the present invention cluster multiple buffer pool system Figure.Wherein, this cluster multiple buffer pool system includes Centroid 1000, overall situation Buffer Pool and calculates node, This overall situation Buffer Pool includes the buffer joint independent of this Centroid.Centroid 1000 comprises the steps that Receive unit 1001, acquiring unit 1002 and transmitting element 1003.

Receive unit 1001, for receiving the page request that this calculating node sends.

Acquiring unit 1002, the page for the node serial number according to this calculating node and this memory pages is compiled Number, obtain this memory pages storage positional information in this overall situation Buffer Pool.

Transmitting element 1003, for sending this memory pages in this overall situation Buffer Pool to this calculating node Storage positional information.

Alternatively, Centroid 1000 also includes administrative unit 1004, is used for managing this overall situation Buffer Pool Buffer joint.

Alternatively, as an embodiment, administrative unit 1004 specifically for: by receiving unit 1001 Receiving the registration information of the first buffer joint, this registration information carries this first buffer joint Node serial number and memory source information, wherein, this first buffer joint is the buffering outside this overall situation Buffer Pool Node；The memory source of this first buffer joint is noted by the memory source information according to this first buffer joint Volume is in this overall situation Buffer Pool.

Alternatively, as another embodiment, administrative unit 1004 specifically for: by receiving unit 1001 receive second buffer joints send de-registration request information, this de-registration request information carry this second delay Rushing the node serial number of node, wherein, this second buffer joint is the buffer joint in this overall situation Buffer Pool； According to this de-registration request information by the Data Migration of this second buffer joint to this overall situation Buffer Pool this In buffer joint beyond two buffer joints；By the memory source information of this second buffer joint from this overall situation Buffer Pool is deleted.

Alternatively, as further embodiment, administrative unit 1004 specifically for: if receiving unit 1001 receive the notice about the second buffer joint fault, then according to this notice by this second buffer joint Data Migration in the buffer joint beyond this second buffer joint in this overall situation Buffer Pool, and by this The memory source information of two buffer joints is deleted from this overall situation Buffer Pool, wherein, and this second buffer joint For the buffer joint in this overall situation Buffer Pool.

Alternatively, as further embodiment, administrative unit 1004 specifically for: by receiving unit 1001 receive the replacement request information that the second buffer joint sends, and this replacement request information carries the 3rd buffering The memory source information of node, wherein, this second buffer joint is the buffer joint in this overall situation Buffer Pool, 3rd buffer joint is the buffer joint outside this overall situation Buffer Pool；According to this replacement request information by this The Data Migration of two buffer joints is in the 3rd buffer joint, and deletes the registration of this second buffer joint Information, and according to the memory source information of the 3rd buffer joint by the memory source of the 3rd buffer joint It is registered to this overall situation Buffer Pool.

It addition, the method that Centroid 1000 can also carry out Fig. 7, and possesses Centroid at Fig. 2 extremely Function in embodiment illustrated in fig. 6, implements and refers to Fig. 2 to embodiment illustrated in fig. 6, the present invention Embodiment does not repeats them here.

Figure 11 is the structural representation calculating node 1100 of embodiment of the present invention cluster multiple buffer pool system Figure.Wherein, this cluster multiple buffer pool system includes Centroid, overall situation Buffer Pool and calculates node 1100, This overall situation Buffer Pool includes the buffer joint independent of this Centroid.Calculate node 1100 to comprise the steps that Transmitting element 1101, reception unit 1102 and access unit 1103.

Transmitting element 1101, for sending page request to this Centroid, this page request carries calculating The page number of the memory pages that the node serial number of node 1100 and calculating node 1100 application access；

Receive unit 1102, for receiving this memory pages of this Centroid transmission at this overall situation Buffer Pool In storage positional information, wherein, this memory pages storage positional information in this overall situation Buffer Pool by This Centroid is complete from this according to the page number of the node serial number and this memory pages that calculate node 1100 Office's Buffer Pool obtains, and this overall situation Buffer Pool is for storing the globally shared buffering of this cluster multiple buffer pool system Data；

Access unit 1103, for believing according to this memory pages storage position in this overall situation Buffer Pool Breath, conducts interviews to this memory pages.

Further, this access unit 1103 specifically for: this memory pages is being read out access Time, select one of this memory pages storage position of at least two buffer joint in this overall situation Buffer Pool It is read out this memory pages accessing；Or when this memory pages is carried out write-access, selecting should The storage position of memory pages at least two buffer joint in this overall situation Buffer Pool is to this memory pages Carry out write-access.

It addition, calculate the method that node 1100 can also carry out Fig. 8, and possesses calculating node in Fig. 2 institute Showing the function in embodiment, implement and refer to embodiment illustrated in fig. 2, the embodiment of the present invention is at this not Repeat again.

Figure 12 is the structural representation of the Centroid 1200 of embodiment of the present invention cluster multiple buffer pool system Figure.Wherein, this cluster multiple buffer pool system includes Centroid 1200, overall situation Buffer Pool and calculates node, This overall situation Buffer Pool includes the buffer joint independent of this Centroid.Centroid 1200 comprises the steps that Communication interface 1201, processor 1202 and memory 1203.

Communication interface 1201, processor 1202 are interconnected by bus 1204 system mutually with memory 1203 Connect.Bus 1204 can be isa bus, pci bus or eisa bus etc..Described bus can be divided For address bus, data/address bus, control bus etc..For ease of represent, in Figure 12 only with one two-way Arrow represents, it is not intended that an only bus or a type of bus.

Memory 1203, is used for depositing program.Specifically, program can include program code, described journey Sequence code includes computer-managed instruction.Memory 1203 can include read-only storage and arbitrary access Memory, and provide instruction and data to processor 1202.Memory 1203 may comprise high-speed RAM Memory, it is also possible to also include nonvolatile memory (non-volatile memory), for example, at least one Individual magnetic disc store.

Communication interface 1201, can be used for receiving the page request that this calculating node sends.

Communication interface 1201 can be additionally used in and sends this memory pages in overall situation Buffer Pool to this calculating node Storage positional information.Wherein, this overall situation Buffer Pool is used for storing this cluster many buffering of multiple buffer pool cluster The globally shared buffered data of cell system, this memory pages storage positional information in this overall situation Buffer Pool For this calculating node, this memory pages is conducted interviews.

Processor 1202, is performed the program that memory 1203 is deposited, is received by communication interface 1201 The page request that this calculating node sends, according to node serial number and the page of this memory pages of this calculating node Face is numbered, and obtains this memory pages storage positional information in this overall situation Buffer Pool, and is connect by communication Mouth 1201 sends this memory pages storage positional information in overall situation Buffer Pool to this calculating node.

The above-mentioned method performed such as the Centroid disclosed in Fig. 2 to Fig. 7 any embodiment of the present invention can be answered In processor 1202, or realized by processor 1202.Processor 1202 is probably a kind of collection Become circuit chip, there is the disposal ability of signal.During realizing, each step of said method is permissible Completed by the integrated logic circuit of the hardware in processor 1202 or the instruction of software form.Above-mentioned Processor 1202 can be general processor, including central processing unit (Central Processing Unit, Be called for short CPU), network processing unit (Network Processor, be called for short NP) etc.；Can also is that numeral Signal processor (DSP), special IC (ASIC), ready-made programmable gate array (FPGA) Or other PLDs, discrete gate or transistor logic, discrete hardware components.Can To realize or to perform disclosed each method, step and the logic diagram in the embodiment of the present invention.General place The reason processor etc. that device can be microprocessor or this processor can also be any routine.In conjunction with this The step of the method disclosed in bright embodiment can be embodied directly in hardware decoding processor and perform, or Hardware and software module combination execution in person's decoding processor complete.Software module may be located at random Memory, flash memory, read-only storage, programmable read only memory or electrically erasable programmable memory, In the storage medium that this areas such as register are ripe.This storage medium is positioned at memory 1203, processor 1202 read the information in memory 1203, complete the step of said method in conjunction with its hardware.

Alternatively, processor 1202 can be additionally used in the buffer joint managing this overall situation Buffer Pool.

Alternatively, as an embodiment, at the buffer joint for managing this overall situation Buffer Pool, process Device 1202 specifically for: received the registration information of the first buffer joint by receiver 1201, should Registration information carries node serial number and the memory source information of this first buffer joint, and wherein, this is years old One buffer joint is the buffer joint outside this overall situation Buffer Pool；Memory source according to this first buffer joint The memory source of this first buffer joint is registered in this overall situation Buffer Pool by information.

Alternatively, as another embodiment, at the buffer joint for managing this overall situation Buffer Pool, place Reason device 1202 specifically for: by receiver 1201 receive second buffer joint send de-registration request believe Breath, this de-registration request information carries the node serial number of this second buffer joint, wherein, this second buffering joint Point is the buffer joint in this overall situation Buffer Pool；According to this de-registration request information by this second buffer joint Data Migration is in the buffer joint beyond this second buffer joint in this overall situation Buffer Pool；This second is delayed The memory source information rushing node is deleted from this overall situation Buffer Pool.

Alternatively, as further embodiment, at the buffer joint for managing this overall situation Buffer Pool, place Reason device 1202 specifically for: if receiver 1201 receives the notice about the second buffer joint fault, Then according to this notice by this second buffering in the Data Migration of this second buffer joint to this overall situation Buffer Pool In buffer joint beyond node, and the memory source information of this second buffer joint is buffered from this overall situation Deleting in pond, wherein, this second buffer joint is the buffer joint in this overall situation Buffer Pool.

Alternatively, as further embodiment, at the buffer joint for managing this overall situation Buffer Pool, place Reason device 1202 specifically for: by receiver 1201 receive second buffer joint send replacement request believe Breath, this replacement request information carries the memory source information of the 3rd buffer joint, wherein, this second buffering Node is the buffer joint in this overall situation Buffer Pool, and the 3rd buffer joint is delaying outside this overall situation Buffer Pool Rush node；According to this replacement request information by the Data Migration of this second buffer joint to the 3rd buffering joint In point, and delete the log-on message of this second buffer joint, and provide according to the internal memory of the 3rd buffer joint The memory source of the 3rd buffer joint is registered to this overall situation Buffer Pool by source information.

It addition, the method that Centroid 1200 can also carry out Fig. 7, and possesses Centroid at Fig. 2 extremely Function in embodiment illustrated in fig. 6, implements and refers to Fig. 2 to embodiment illustrated in fig. 7, the present invention Embodiment does not repeats them here.

Figure 13 is the structural representation calculating node 1300 of embodiment of the present invention cluster multiple buffer pool system Figure.Wherein, this cluster multiple buffer pool system includes Centroid, overall situation Buffer Pool and calculates node 1300, This overall situation Buffer Pool includes the buffer joint independent of this Centroid.Calculate node 1300 to comprise the steps that Communication interface 1301, processor 1302 and memory 1303.

Communication interface 1301, processor 1302 are interconnected by bus 1304 system mutually with memory 1303 Connect.Bus 1304 can be isa bus, pci bus or eisa bus etc..Described bus can be divided For address bus, data/address bus, control bus etc..For ease of represent, in Figure 13 only with one two-way Arrow represents, it is not intended that an only bus or a type of bus.

Memory 1303, is used for depositing program.Specifically, program can include program code, described journey Sequence code includes computer-managed instruction.Memory 1303 can include read-only storage and arbitrary access Memory, and provide instruction and data to processor 1302.Memory 1303 may comprise high-speed RAM Memory, it is also possible to also include nonvolatile memory (non-volatile memory), for example, at least one Individual magnetic disc store.

Communication interface 1301, for sending page request to this Centroid.

Wherein, this page request is carried to calculate the node serial number of node 1300 and calculate node 1300 and is applied for The page number of the memory pages accessed.

Communication interface 1301 can be additionally used in this memory pages of reception this Centroid transmission and delays in this overall situation Rush the storage positional information in pond.

Wherein, this memory pages storage positional information in this overall situation Buffer Pool by this Centroid according to Calculate the node serial number of node 1300 and the page number of this memory pages obtain from this overall situation Buffer Pool, This overall situation Buffer Pool is for storing the globally shared buffered data of this cluster multiple buffer pool system；

Processor 1303, performs the program that memory 1303 is deposited, by communication interface 1301 to this Centroid sends page request, and by communication interface 1301 receive that this Centroid sends this is interior Deposit page storage positional information in this overall situation Buffer Pool, and buffer in this overall situation according to this memory pages Storage positional information in pond, conducts interviews to this memory pages.

The above-mentioned method performed such as the calculating node disclosed in Fig. 2, Fig. 8 any embodiment of the present invention can be answered In processor 1302, or realized by processor 1302.Processor 1302 is probably a kind of collection Become circuit chip, there is the disposal ability of signal.During realizing, each step of said method is permissible Completed by the integrated logic circuit of the hardware in processor 1302 or the instruction of software form.Above-mentioned Processor 1302 can be general processor, including central processing unit (Central Processing Unit, Be called for short CPU), network processing unit (Network Processor, be called for short NP) etc.；Can also is that numeral Signal processor (DSP), special IC (ASIC), ready-made programmable gate array (FPGA) Or other PLDs, discrete gate or transistor logic, discrete hardware components.Can To realize or to perform disclosed each method, step and the logic diagram in the embodiment of the present invention.General place The reason processor etc. that device can be microprocessor or this processor can also be any routine.In conjunction with this The step of the method disclosed in bright embodiment can be embodied directly in hardware decoding processor and perform, or Hardware and software module combination execution in person's decoding processor complete.Software module may be located at random Memory, flash memory, read-only storage, programmable read only memory or electrically erasable programmable memory, In the storage medium that this areas such as register are ripe.This storage medium is positioned at memory 1303, processor 1302 read the information in memory 1303, complete the step of said method in conjunction with its hardware.

Further, this access unit 1303 specifically for: this memory pages is being read out access Time, select one of this memory pages storage position of at least two buffer joint in this overall situation Buffer Pool It is read out this memory pages accessing；Or when this memory pages is carried out write-access, selecting should The storage position of memory pages at least two buffer joint in this overall situation Buffer Pool is to this memory pages Carry out write-access.

It addition, calculate the method that node 1300 can also carry out Fig. 8, and possesses calculating node in Fig. 2 institute Show the function in embodiment, implement and refer to Fig. 2, embodiment illustrated in fig. 8, the embodiment of the present invention Do not repeat them here.

Those of ordinary skill in the art are it is to be appreciated that combine each of the embodiments described herein description The unit of example and algorithm steps, it is possible to electronic hardware or computer software and the knot of electronic hardware Incompatible realization.These functions perform with hardware or software mode actually, depend on the spy of technical scheme Fixed application and design constraint.Professional and technical personnel can use not Tongfang to each specifically should being used for Method realizes described function, but this realization is it is not considered that beyond the scope of this invention.

Those skilled in the art is it can be understood that arrive, and for convenience and simplicity of description, above-mentioned retouches The specific works process of system, device and the unit stated, is referred to the correspondence in preceding method embodiment Process, does not repeats them here.

In several embodiments provided herein, it should be understood that disclosed system, device and Method, can realize by another way.Such as, device embodiment described above is only shown Meaning property, such as, the division of described unit, be only a kind of logic function and divide, actual can when realizing There to be other dividing mode, the most multiple unit or assembly can in conjunction with or be desirably integrated into another System, or some features can ignore, or do not perform.Another point, shown or discussed each other Coupling direct-coupling or communication connection can be the INDIRECT COUPLING by some interfaces, device or unit Or communication connection, can be electrical, machinery or other form.

The described unit illustrated as separating component can be or may not be physically separate, makees The parts shown for unit can be or may not be physical location, i.e. may be located at a place, Or can also be distributed on multiple NE.Can select according to the actual needs part therein or The whole unit of person realizes the purpose of the present embodiment scheme.

It addition, each functional unit in each embodiment of the present invention can be integrated in a processing unit In, it is also possible to it is that unit is individually physically present, it is also possible to two or more unit are integrated in one In individual unit.

If described function realizes using the form of SFU software functional unit and as independent production marketing or make Used time, can be stored in a computer read/write memory medium.Based on such understanding, the present invention The part that the most in other words prior art contributed of technical scheme or the portion of this technical scheme Dividing and can embody with the form of software product, this computer software product is stored in a storage medium In, including some instructions with so that computer equipment (can be personal computer, server, Or the network equipment etc.) perform all or part of step of method described in each embodiment of the present invention.And it is front The storage medium stated includes: USB flash disk, portable hard drive, read-only storage (ROM, Read-Only Memory), Random access memory (RAM, Random Access Memory), magnetic disc or CD etc. are various can To store the medium of program code.

The above, the only detailed description of the invention of the present invention, but protection scope of the present invention is not limited to In this, any those familiar with the art, can be easily in the technical scope that the invention discloses Expect change or replace, all should contain within protection scope of the present invention.Therefore, the protection of the present invention Scope should described be as the criterion with scope of the claims.

Claims

1. a cluster multiple buffer pool system, it is characterised in that including: Centroid, overall situation buffering Pond and calculating node, wherein

Described overall situation Buffer Pool is for storing the globally shared buffered data of described system, and the described overall situation buffers Pond includes the buffer joint independent of described Centroid；

Described calculating node is for sending page request to described Centroid, and described page request carries institute State the node serial number of calculating node and the page number of the memory pages of described calculating node application access；

Described Centroid is for the node serial number according to described calculating node and the page of described memory pages Face is numbered, and provides described memory pages storage position in described overall situation Buffer Pool to described calculating node Information；

Described calculating node be additionally operable to according to described memory pages described the overall situation Buffer Pool in storage position Confidence ceases, and conducts interviews described memory pages.

2. the system as claimed in claim 1, it is characterised in that described Centroid is additionally operable to management The buffer joint of described overall situation Buffer Pool.

3. system as claimed in claim 2, it is characterised in that manage described overall situation buffering being used for The buffer joint in pond, described Centroid specifically for: receive the registration information of the first buffer joint, Described registration information carries node serial number and the memory source information of described first buffer joint, its In, described first buffer joint is the buffer joint outside described overall situation Buffer Pool；According to described first buffering The memory source of described first buffer joint is registered to described overall situation buffering by the memory source information of node Chi Zhong.

4. system as claimed in claim 3, it is characterised in that described memory source information includes institute State the memory address of the first buffer joint, internal memory length and remote direct memory and access RDMA access pass Key word KEY.

5. system as claimed in claim 2, it is characterised in that manage described overall situation buffering being used for The buffer joint in pond, described Centroid specifically for: receive second buffer joint send de-registration request Information, described de-registration request information carries the node serial number of described second buffer joint, wherein, described Two buffer joints are the buffer joint in described overall situation Buffer Pool；According to described de-registration request information by described The Data Migration of the second buffer joint is to slow beyond the second buffer joint described in described overall situation Buffer Pool Rush in node；The memory source information of described second buffer joint is deleted from described overall situation Buffer Pool.

6. system as claimed in claim 2, it is characterised in that manage described overall situation buffering being used for The buffer joint in pond, described Centroid specifically for: if received about the second buffer joint fault Notice, then according to described notice by the Data Migration of described second buffer joint to described overall situation Buffer Pool In buffer joint beyond described second buffer joint, and the memory source of described second buffer joint is believed Cease and delete from described overall situation Buffer Pool, wherein, in described second buffer joint is described overall situation Buffer Pool Buffer joint.

7. system as claimed in claim 2, it is characterised in that manage described overall situation buffering being used for The buffer joint in pond, described Centroid specifically for: receive second buffer joint send replacement request Information, described replacement request information carries the memory source information of the 3rd buffer joint, wherein, described Two buffer joints are the buffer joint in described overall situation Buffer Pool, and described 3rd buffer joint is the described overall situation Buffer joint outside Buffer Pool；According to described replacement request information, the data of described second buffer joint are moved Move on in described 3rd buffer joint, and delete the log-on message of described second buffer joint, and according to institute The memory source of described 3rd buffer joint is registered to institute by the memory source information stating the 3rd buffer joint State overall situation Buffer Pool.

8. the system as described in any one of claim 1 to 7, it is characterised in that described overall situation buffering Pond includes at least 2 buffer joints, stores institute in all buffer joints in described overall situation Buffer Pool State at least 2 backups of the globally shared buffered data of overall situation Buffer Pool.

9. system as claimed in claim 8, it is characterised in that described memory pages is in the described overall situation Storage positional information in Buffer Pool includes described memory pages at least two in described overall situation Buffer Pool The storage position of individual buffer joint, described calculating node according to described memory pages described the overall situation Buffer Pool In storage positional information described memory pages conducted interviews include:

When described memory pages is read out accessing by described calculating node, described calculating node selects institute State a pair institute of the storage position of memory pages at least two buffer joint in described overall situation Buffer Pool State memory pages to be read out accessing；Or

When described calculating node carries out write-access to described memory pages, described calculating node selects institute State the storage position of memory pages at least two buffer joint in described overall situation Buffer Pool in described Deposit the page and carry out write-access.

10. the system as claimed in claim 1, it is characterised in that described page request is also carried described The page byte number of memory pages, described page byte number combines described page for described Centroid The page number in face determines described storage positional information.

The overall Buffer Pool management method of 11. 1 kinds of cluster multiple buffer pool systems, described cluster multiple buffer pool System includes Centroid, overall situation Buffer Pool and calculates node, and described overall situation Buffer Pool includes independent of institute Stating the buffer joint of Centroid, described method includes:

Described calculating node sends page request to described Centroid, and described page request carries described meter The page number of the memory pages that the node serial number of operator node and described calculating node application access；

Described Centroid is compiled according to the node serial number of described calculating node and the page of described memory pages Number, provide described memory pages storage position letter in described overall situation Buffer Pool to described calculating node Breath, wherein, described overall situation Buffer Pool is for storing the globally shared buffering of described cluster multiple buffer pool system Data；

Described calculating node according to described memory pages described the overall situation Buffer Pool in storage positional information, Described memory pages is conducted interviews.

12. methods as claimed in claim 11, it is characterised in that described method also includes: described The buffer joint of Centroid management described overall situation Buffer Pool.

13. methods as claimed in claim 12, it is characterised in that the management of described Centroid is described The buffer joint of overall situation Buffer Pool includes: described Centroid receives the registration request letter of the first buffer joint Ceasing, described registration information carries node serial number and the memory source information of described first buffer joint, Wherein, the buffer joint outside described first buffer joint is described overall situation Buffer Pool；Described Centroid root According to the memory source information of described first buffer joint, the memory source of described first buffer joint is registered In described overall situation Buffer Pool.

14. methods as claimed in claim 12, it is characterised in that the management of described Centroid is described The buffer joint of overall situation Buffer Pool includes: described Centroid receives the cancellation of the second buffer joint transmission please Information, described de-registration request information is asked to carry the node serial number of described second buffer joint, wherein, described Second buffer joint is the buffer joint in described overall situation Buffer Pool；Described Centroid is according to described cancellation Solicited message is delayed described in the Data Migration of described second buffer joint to described overall situation Buffer Pool second Rush in the buffer joint beyond node；The memory source of described second buffer joint is believed by described Centroid Cease and delete from described overall situation Buffer Pool.

15. methods as claimed in claim 12, it is characterised in that the management of described Centroid is described The buffer joint of overall situation Buffer Pool includes: if described Centroid receives about the second buffer joint event The notice of barrier, the Data Migration of described second buffer joint is arrived by the most described Centroid according to described notice In buffer joint beyond second buffer joint described in described overall situation Buffer Pool, and will described second buffer The memory source information of node is deleted from described overall situation Buffer Pool, and wherein, described second buffer joint is Buffer joint in described overall situation Buffer Pool.

16. methods as claimed in claim 12, it is characterised in that the management of described Centroid is described The buffer joint of overall situation Buffer Pool includes: described Centroid receives the replacing of the second buffer joint transmission please Information, described replacement request information is asked to carry the memory source information of the 3rd buffer joint, wherein, described Second buffer joint be described the overall situation Buffer Pool in buffer joint, described 3rd buffer joint be described entirely Buffer joint outside office's Buffer Pool；Described Centroid delays described second according to described replacement request information Rush the Data Migration of node in described 3rd buffer joint, and delete the registration of described second buffer joint Information, and according to the memory source information of described 3rd buffer joint by the internal memory of described 3rd buffer joint Resource registering is to described overall situation Buffer Pool.

17. methods as described in any one of claim 11 to 16, it is characterised in that the described overall situation is delayed Rush pond and include at least 2 buffer joints, all buffer joints in described overall situation Buffer Pool store At least 2 backups of the globally shared buffered data of described overall situation Buffer Pool.

18. methods as claimed in claim 17, it is characterised in that described memory pages described entirely Storage positional information in office's Buffer Pool include described memory pages in described overall Buffer Pool at least The storage position of two buffer joints, described calculating node buffers in the described overall situation according to described memory pages Described memory pages is conducted interviews and includes by the storage positional information in pond:

The overall Buffer Pool management method of 19. 1 kinds of cluster multiple buffer pool systems, described cluster multiple buffer pool System includes Centroid, overall situation Buffer Pool and calculates node, and described overall situation Buffer Pool includes independent of institute Stating the buffer joint of Centroid, described method includes:

Described Centroid receives the page request that described calculating node sends, and described page request carries institute State the node serial number of calculating node and the page number of the memory pages of described calculating node application access；

Described Centroid is compiled according to the node serial number of described calculating node and the page of described memory pages Number, obtain described memory pages described the overall situation Buffer Pool in storage positional information, wherein, described entirely Office's Buffer Pool is for storing the globally shared buffered data of described cluster multiple buffer pool system；

Described Centroid sends described memory pages in described overall situation Buffer Pool to described calculating node Storage positional information, wherein, described memory pages described the overall situation Buffer Pool in storage positional information For described calculating node, described memory pages is conducted interviews.

20. methods as claimed in claim 19, it is characterised in that described method also includes: described The buffer joint of Centroid management described overall situation Buffer Pool.

21. methods as claimed in claim 20, it is characterised in that the management of described Centroid is described The buffer joint of overall situation Buffer Pool includes: described Centroid receives the registration request letter of the first buffer joint Ceasing, described registration information carries node serial number and the memory source information of described first buffer joint, Wherein, the buffer joint outside described first buffer joint is described overall situation Buffer Pool；Described Centroid root According to the memory source information of described first buffer joint, the memory source of described first buffer joint is registered In described overall situation Buffer Pool.

22. methods as claimed in claim 20, it is characterised in that the management of described Centroid is described The buffer joint of overall situation Buffer Pool includes: described Centroid receives the cancellation of the second buffer joint transmission please Information, described de-registration request information is asked to carry the node serial number of described second buffer joint, wherein, described Second buffer joint is the buffer joint in described overall situation Buffer Pool；Described Centroid is according to described cancellation Solicited message is delayed described in the Data Migration of described second buffer joint to described overall situation Buffer Pool second Rush in the buffer joint beyond node；The memory source of described second buffer joint is believed by described Centroid Cease and delete from described overall situation Buffer Pool.

23. methods as claimed in claim 20, it is characterised in that the management of described Centroid is described The buffer joint of overall situation Buffer Pool includes: if described Centroid receives about the second buffer joint event The notice of barrier, the Data Migration of described second buffer joint is arrived by the most described Centroid according to described notice In buffer joint beyond second buffer joint described in described overall situation Buffer Pool, and will described second buffer The memory source information of node is deleted from described overall situation Buffer Pool, and wherein, described second buffer joint is Buffer joint in described overall situation Buffer Pool.

24. methods as claimed in claim 20, it is characterised in that the management of described Centroid is described The buffer joint of overall situation Buffer Pool includes: described Centroid receives the replacing of the second buffer joint transmission please Information, described replacement request information is asked to carry the memory source information of the 3rd buffer joint, wherein, described Second buffer joint be described the overall situation Buffer Pool in buffer joint, described 3rd buffer joint be described entirely Buffer joint outside office's Buffer Pool；Described Centroid delays described second according to described replacement request information Rush the Data Migration of node in described 3rd buffer joint, and delete the registration of described second buffer joint Information, and according to the memory source information of described 3rd buffer joint by the internal memory of described 3rd buffer joint Resource registering is to described overall situation Buffer Pool.

25. methods as described in any one of claim 19 to 24, it is characterised in that the described overall situation is delayed Rush pond and include at least 2 buffer joints, all buffer joints in described overall situation Buffer Pool store At least 2 backups of the globally shared buffered data of described overall situation Buffer Pool, described memory pages is described Storage positional information in overall situation Buffer Pool include described memory pages in described overall Buffer Pool extremely The storage position of few two buffer joints.

The overall Buffer Pool management method of 26. 1 kinds of cluster multiple buffer pool systems, described cluster multiple buffer pool System includes Centroid, overall situation Buffer Pool and calculates node, and described overall situation Buffer Pool includes independent of institute Stating the buffer joint of Centroid, described method includes:

Described calculating node receives the described memory pages of described Centroid transmission in described overall situation buffering Storage positional information in pond, wherein, described memory pages storage position in described overall situation Buffer Pool Information by described Centroid according to the node serial number of described calculating node and the page of described memory pages Numbering and obtain from described overall situation Buffer Pool, described overall situation Buffer Pool is used for storing described cluster multiple buffer pool system The globally shared buffered data of system；

27. methods as claimed in claim 26, it is characterised in that described overall situation Buffer Pool includes At least 2 buffer joints, store the described overall situation in all buffer joints in described overall situation Buffer Pool and delay Rushing at least 2 backups of the globally shared buffered data in pond, described memory pages is at described overall situation Buffer Pool In storage positional information include described memory pages described the overall situation Buffer Pool at least two buffering The storage position of node, described calculating node is according to described memory pages depositing in described overall situation Buffer Pool Storage positional information, conducts interviews to described memory pages and includes:

The Centroid of 28. 1 kinds of cluster multiple buffer pool systems, it is characterised in that the many bufferings of described cluster Cell system includes described Centroid, overall situation Buffer Pool and calculates node, and described overall situation Buffer Pool includes solely Standing on the buffer joint of described Centroid, described Centroid includes:

Receiving unit, for receiving the page request that described calculating node sends, described page request is carried The page number of the memory pages that the node serial number of described calculating node and described calculating node application access；

Acquiring unit, the page for the node serial number according to described calculating node and described memory pages is compiled Number, obtain described memory pages described the overall situation Buffer Pool in storage positional information, wherein, described entirely Office's Buffer Pool is for storing the globally shared buffered data of described cluster multiple buffer pool system；

Transmitting element, for sending described memory pages in described overall situation Buffer Pool to described calculating node Storage positional information, wherein, described memory pages described the overall situation Buffer Pool in storage positional information For described calculating node, described memory pages is conducted interviews.

29. Centroids as claimed in claim 28, it is characterised in that described Centroid also wraps Include administrative unit, for managing the buffer joint of described overall situation Buffer Pool.

30. Centroids as claimed in claim 29, it is characterised in that described administrative unit is concrete For:

Received the registration information of the first buffer joint by described reception unit, described registration request is believed Breath carries node serial number and the memory source information of described first buffer joint, wherein, described first buffering Node is the buffer joint outside described overall situation Buffer Pool；

The internal memory of described first buffer joint is provided by the memory source information according to described first buffer joint Login to described the overall situation Buffer Pool in.

31. Centroids as claimed in claim 29, it is characterised in that described administrative unit is concrete For:

Receiving, by described reception unit, the de-registration request information that the second buffer joint sends, described cancellation please Asking information to carry the node serial number of described second buffer joint, wherein, described second buffer joint is described Buffer joint in overall situation Buffer Pool；

According to described de-registration request information, the Data Migration of described second buffer joint is delayed to the described overall situation Rush described in pond in the buffer joint beyond the second buffer joint；

The memory source information of described second buffer joint is deleted from described overall situation Buffer Pool.

32. Centroids as claimed in claim 29, it is characterised in that described administrative unit is concrete For:

If described reception unit receives the notice about the second buffer joint fault, then according to described logical Know the second buffer joint described in the Data Migration of described second buffer joint to described overall situation Buffer Pool In buffer joint in addition, and by the memory source information of described second buffer joint from described overall situation buffering Deleting in pond, wherein, described second buffer joint is the buffer joint in described overall situation Buffer Pool.

33. Centroids as claimed in claim 29, it is characterised in that described administrative unit is concrete For:

Receiving, by described reception unit, the replacement request information that the second buffer joint sends, described replacing please Asking information to carry the memory source information of the 3rd buffer joint, wherein, described second buffer joint is described Buffer joint in overall situation Buffer Pool, described 3rd buffer joint is the buffering joint outside described overall situation Buffer Pool Point；

According to described replacement request information, the Data Migration of described second buffer joint is delayed to the described 3rd Rush in node, and delete the log-on message of described second buffer joint, and according to described 3rd buffer joint Memory source information the memory source of described 3rd buffer joint is registered to described the overall situation Buffer Pool.

34. Centroids as described in any one of claim 28 to 33, it is characterised in that described entirely Office's Buffer Pool includes at least 2 buffer joints, deposits in all buffer joints in described overall situation Buffer Pool Storing up at least 2 backups of the globally shared buffered data of described overall situation Buffer Pool, described memory pages exists Storage positional information in described overall situation Buffer Pool includes that described memory pages is in described overall situation Buffer Pool The storage position of at least two buffer joint.

The calculating node of 35. 1 kinds of cluster multiple buffer pool systems, it is characterised in that the many bufferings of described cluster Cell system include Centroid, the overall situation Buffer Pool and calculate node, described the overall situation Buffer Pool include independent of The buffer joint of described Centroid, described calculating node includes:

Transmitting element, for sending page request to described Centroid, described page request is carried described Calculate the node serial number of node and the page number of the memory pages of described calculating node application access；

Receiving unit, the described memory pages sent for receiving described Centroid buffers in the described overall situation Storage positional information in pond, wherein, described memory pages storage position in described overall situation Buffer Pool Information by described Centroid according to the node serial number of described calculating node and the page of described memory pages Numbering and obtain from described overall situation Buffer Pool, described overall situation Buffer Pool is used for storing described cluster multiple buffer pool system The globally shared buffered data of system；

Access unit, for according to described memory pages described the overall situation Buffer Pool in storage position letter Breath, conducts interviews to described memory pages.

36. calculate node as claimed in claim 35, it is characterised in that in described overall situation Buffer Pool Including at least 2 buffer joints, all buffer joints in described overall situation Buffer Pool store described entirely At least 2 backups of the globally shared buffered data of office's Buffer Pool, described memory pages delays in the described overall situation Rush the storage positional information in pond and include described memory pages at least two in described overall situation Buffer Pool The storage position of buffer joint, described access unit specifically for:

When being read out described memory pages accessing, select described memory pages in described overall situation buffering Memory pages described in a pair of the storage position of at least two buffer joint in pond is read out accessing；Or Person

When described memory pages is carried out write-access, select described memory pages in described overall situation buffering The storage position of at least two buffer joint in pond carries out write-access to described memory pages.