CN104216758B - A kind of read-write operation performance optimization method and device - Google Patents
A kind of read-write operation performance optimization method and device Download PDFInfo
- Publication number
- CN104216758B CN104216758B CN201410419211.9A CN201410419211A CN104216758B CN 104216758 B CN104216758 B CN 104216758B CN 201410419211 A CN201410419211 A CN 201410419211A CN 104216758 B CN104216758 B CN 104216758B
- Authority
- CN
- China
- Prior art keywords
- virtual machine
- mark
- processor
- read
- module
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
Landscapes
- Memory System Of A Hierarchy Structure (AREA)
Abstract
The invention discloses a kind of read-write operation performance optimization method and device.The described method includes:The read-write requests that the processor in processor cluster is sent are received, wherein, read-write requests contain the mark of processor;With the foundation that is identified as of processor, query configuration table to obtain the mark for the virtual machine that processor is currently running, wherein, the mapping relations of the mark for the virtual machine that allocation list contains the mark of processor and processor is currently running;The mark of the virtual machine is sent to slave device, the mark for the virtual machine being currently running according to processor carries out performance optimization processing to read-write operation.Such scheme can know the mark for the virtual machine for sending read-write requests, and read-write operation is optimized according to the mark of virtual machine.
Description
Technical field
The present invention relates to processor, more particularly to a kind of read-write operation performance optimization method and device.
Background technology
Virtualization technology is a critically important technology in computer technology, can be in same meter by virtualization technology
Multiple virtual machines are run on calculation machine hardware at the same time, an operating system is run on each virtual machine.Each operating system can phase
Mutually operate independently from the application program of oneself so that do not perceived mutually between operating system.But when the quantity of virtual machine compares
When more, the performance that different virtual machine is written and read cache buffer (Cache) operation will influence each other, because
Cache needs to handle the read-write requests of multiple virtual machines at the same time, and the access of these virtual machines differs from one another.It is for example, virtual
The size of distribution of the application program run in machine A to Cache is very sensitive, and more Cache is assigned to if virtual machine A,
Then the performance for the read-write operation that virtual machine A is showed is better;Distribution of the application program run in virtual machine B to Cache
Size it is insensitive, even if for virtual machine B distribution Cache it is more again, the read-write operation performance that virtual machine B is showed is still
Any change is not had, or even the Cache for thus giving virtual machine B on the contrary is more, causes to distribute to as virtual machine A is this
Tail off, influenced as this Cache to distribution of virtual machine A are big to the Cache of the virtual machine of the Cache magnitudes of distribution
The read-write operation performance of small sensitive virtual machine.
So if can recognize that virtual machine, the characteristics of read-write requests sent according to virtual machine, is carried out at performance optimization
Reason, then can greatly improve the performance that operation is written and read to Cache.But it in the prior art, can only at most identify reading
Write request is that some processor is sent.However, on the one hand for small virtual machine, virtual machine can in a upper period
It can operate in processor 1, may be operated in processor 2 in next period, on the other hand, for large-scale virtual machine
For, virtual machine may operate in multiple processors at the same time.So the prior art cannot identify send read-write requests be
Which virtual machine.
The content of the invention
Technical problem to be solved of the embodiment of the present invention is, there is provided a kind of virtual machine sign acquiring method, device with
And equipment, it can know the mark for the virtual machine for sending read-write requests, read-write operation is optimized according to the mark of virtual machine.
The application first aspect provides a kind of read-write operation performance optimization method, the described method includes:Receive positioned at place
The read-write requests that the processor in device cluster is sent are managed, wherein, the read-write requests contain the mark of the processor;With described
Processor is identified as foundation, query configuration table to obtain the mark for the virtual machine that the processor is currently running, wherein,
The mapping of the mark for the virtual machine that the mark that the allocation list contains the processor is currently running with the processor
Relation;The mark of the virtual machine is sent to slave device so that the slave device is currently running according to the processor
The mark of virtual machine carries out performance optimization processing to read-write operation.
With reference to first aspect, in the first possible embodiment of the application first aspect, it is located at processor receiving
Before the read-write requests that processor in cluster is sent, the method further includes:The processor is obtained by operating system to send
The mark of virtual machine that is currently running of the processor;Establish the mark of the processor and the processor currently just
Mapping relations between the mark of the virtual machine of operation;The mark of the processor and the processor are currently running
Virtual machine mark between mapping relations be loaded into allocation list.
With reference to first aspect, in second of the application first aspect possible embodiment, the slave device is to described
Virtual machine carries out property optimized handling method:It is described virtual according to the acquisition of the mark of the virtual machine by the slave device
Machine is written and read cache buffer the rule of operation, and according to the rule receive next virtual machine read-write please
Before asking, data are got into cache buffer in advance.
With reference to first aspect, in the third possible embodiment of the application first aspect, as slave device according to
The mark for the virtual machine that processor is currently running, which carries out performance optimization processing, to be included:If scanning is run in the virtual machine
Class method, then be limited to the virtual machine distribution cache buffer.
With reference to first aspect, in the 4th kind of possible embodiment of the application first aspect, as slave device according to
The mark for the virtual machine that processor is currently running, which carries out performance optimization processing, to be included:Institute is identified as according to the virtual machine
State virtual machine distribution road, wherein, the road be cache buffer a structural units, at least part virtual machine respectively by
It is assigned to the different roads.
The application second aspect provides a kind of read-write operation performance optimization method, the described method includes:Receive switching model
The mark for the virtual machine that processor transmitted by block is currently running, wherein, the mark of the virtual machine is by described turn
Connection module inquiry is stored in switching model allocation list in the block and obtains, the allocation list contain the mark of the processor with
The mapping relations of the mark for the virtual machine that the processor is currently running;The void being currently running according to the processor
The mark of plan machine carries out performance optimization processing.
With reference to second aspect, in the first possible embodiment of the application second aspect, worked as according to the processor
Before the mark of virtual machine that is currently running carry out performance optimization processing, including:The void is obtained according to the mark of the virtual machine
Plan machine is written and read cache buffer the rule of operation, and is receiving next virtual machine read-write according to the rule
Before request, accessed data are got into cache buffer in advance.
With reference to second aspect, in second of possible embodiment of the application second aspect, worked as according to the processor
Before the mark of virtual machine that is currently running carry out performance optimization processing, including:If operation scanning class method in the virtual machine,
Then it is limited to the virtual machine distribution cache buffer.
With reference to second aspect, in the third possible embodiment of the application second aspect, worked as according to the processor
Before the mark of virtual machine that is currently running carry out performance optimization processing and include:The virtual machine is identified as according to the virtual machine
Road is distributed, wherein, the road is a structural units of cache buffer, and at least part virtual machine is respectively allocated not
The same road.
The application third aspect provides a kind of read-write operation performance optimization device, and described device includes:Receiving module, look into
Module and sending module are ask, the receiving module is used to receive the read-write requests that the processor in processor cluster is sent,
Wherein, the read-write requests contain the mark of the processor, and the mark of the processor is sent to by the receiving module
The enquiry module;The enquiry module is used for the mark for receiving the processor, is identified as foundation with the processor, looks into
Allocation list is ask to obtain the mark for the virtual machine that the processor is currently running, wherein, the allocation list contains described
The mapping relations of the mark for the virtual machine that mark and the processor of processor are currently running, the enquiry module are used for
The mark of the virtual machine is sent to the sending module;The sending module is used for the mark for receiving the virtual machine, to
Slave device sends the mark of the virtual machine so that the virtual machine that the slave device is currently running according to the processor
Mark carries out performance optimization processing to read-write operation.
With reference to the third aspect, in the first possible embodiment of the application third aspect, described device further includes:Obtain
Modulus block, establish module and load-on module, and the acquisition module is used for the institute for obtaining the processor by operating system and sending
State the mark for the virtual machine that processor is currently running, the acquisition module is by the mark of the processor and the virtual machine
Mark be sent to and described establish module;The module of establishing is used to receive the mark of the processor and the mark of the virtual machine
Know, the mapping relations for establishing the mark of the processor between the mark of virtual machine that the processor is currently running,
The mapping relations are sent to the load-on module by the module of establishing;The load-on module is used to receive the mapping pass
It is that the mapping relations between the mark for the virtual machine for being currently running the mark of the processor and the processor load
Into allocation list.
With reference to the third aspect, in second of possible embodiment of the application third aspect, by the slave device according to
The mark of the virtual machine obtains the rule that the virtual machine is written and read cache buffer operation, and according to the rule
Accessed data are got cache buffer by rule in advance before next virtual machine read-write requests are received.
With reference to the third aspect, in the third possible embodiment of the application third aspect, if in the virtual machine
Operation scanning class method, then be limited to the virtual machine by slave device and distribute cache buffer.
With reference to the third aspect, in the 4th kind of possible embodiment of the application third aspect, as slave device according to
Virtual machine is identified as the virtual machine distribution road, wherein, the road is a structural units of cache buffer, at least
Partial virtual machine is respectively allocated the different roads.
The application fourth aspect provides a kind of read-write operation performance optimization device, and described device includes:Receiving module with
And optimization module, the receiving module are used to receive the virtual machine that the processor transmitted by interconnecting module is currently running
Mark, wherein, the mark of the virtual machine is stored in switching model allocation list in the block by interconnecting module inquiry and obtains
, the mark for the virtual machine that the mark that the allocation list contains the processor is currently running with the processor is reflected
Relation is penetrated, the mark of the virtual machine is sent to the optimization module by the receiving module;The optimization module is used for basis
The mark for the virtual machine that the processor is currently running carries out performance optimization processing.
With reference to fourth aspect, in the first possible embodiment of the application fourth aspect, the optimization module is also used
In the rule for being written and read operation to cache buffer according to the mark acquisition virtual machine of the virtual machine, and according to
Accessed data are got cache buffer by the rule in advance before next virtual machine read-write requests are received.
With reference to fourth aspect, in second of possible embodiment of the application fourth aspect, the optimization module is also used
When the operation scanning class method in the virtual machine, the virtual machine distribution cache buffer is limited to.
With reference to fourth aspect, in the third possible embodiment of the application fourth aspect, the optimization module is also used
Road is distributed in the virtual machine that is identified as according to the virtual machine, wherein, the road is a knot of cache buffer
Structure unit, at least part virtual machine are respectively allocated the different roads.
In above-mentioned scheme, after receiving the read-write requests that the processor in processor cluster is sent, according to processor
Mark query configuration table to obtain the mark for the virtual machine that processor is currently running, it is then, current further according to processor
The mark for the virtual machine being currently running optimizes read-write operation.By implementing the embodiment of the present invention, can be matched somebody with somebody according to inquiry
The mark that table knows the virtual machine for sending read-write requests is put, the read-write requests that virtual machine sends are known according to the mark of virtual machine
The characteristics of feature, the read-write requests then sent according to virtual machine, optimizes read-write operation.
Brief description of the drawings
In order to illustrate more clearly about the embodiment of the present invention or technical scheme of the prior art, below will be to embodiment or existing
There is attached drawing needed in technology description to be briefly described, it should be apparent that, drawings in the following description are only this
Some embodiments of invention, for those of ordinary skill in the art, without creative efforts, can be with
Other attached drawings are obtained according to these attached drawings.
Fig. 1 is the flow chart of one embodiment of read-write operation performance optimization method of the present invention;
Fig. 2 is processor and the Organization Chart of slave device in read-write operation performance optimization method of the present invention;
Fig. 3 is the flow chart of another embodiment of read-write operation performance optimization method of the present invention;
Fig. 4 is the flow chart of the another embodiment of read-write operation performance optimization method of the present invention;
Fig. 5 is the structure diagram of read-write operation performance optimization one embodiment of device of the present invention;
Fig. 6 is the structure diagram of read-write operation performance optimization another embodiment of device of the present invention;
Fig. 7 is the structure diagram of the read-write operation performance optimization another embodiment of device of the present invention.
Embodiment
Below in conjunction with the attached drawing in the embodiment of the present invention, the technical solution in the embodiment of the present invention is carried out clear, complete
Site preparation describes, it is clear that described embodiment is only part of the embodiment of the present invention, instead of all the embodiments.It is based on
Embodiment in the present invention, those of ordinary skill in the art are obtained every other without creative efforts
Embodiment, belongs to the scope of protection of the invention.
It should be noted that the term used in embodiments of the present invention is only merely for the mesh of description specific embodiment
, and it is not intended to be limiting the present invention." the one of the embodiment of the present invention and singulative used in the attached claims
Kind ", " described " and "the" are also intended to including most forms, unless context clearly shows that other implications.It is also understood that this
Term "and/or" used herein refers to and comprising any or all possible group associated of list items purpose of one or more
Close.
Referring to Fig. 1, Fig. 1 is the flow chart of one embodiment of read-write operation performance optimization method of the present invention.Present embodiment
Read-write operation performance optimization method include:
110:Receive the read-write requests that the processor in processor cluster is sent.
Specifically, also referring to Fig. 2, as shown in Figure 2 in a multi-processor system, 16 processors are divided into 4
210,4 processor clusters 210 of processor cluster are respectively processor cluster Cluster0~Cluster3.Each processor cluster 210 wraps
It is respectively processor CPU0~CPU3 containing 4 processors, 211,4 processors 211.In interconnecting module 220, for each processing
Device cluster 210 generates an allocation list 222, so, there are totally 4 allocation lists 222 of allocation list 0~3.Allocation list 222 is used for record pair
Answer the mark for the virtual machine that the processor 211 in processor cluster 210 is currently running.Interconnecting module 220 also passes through system bus 230
It is connected with multiple slave devices 240.Wherein, slave device 240 can be Cache etc..Processor cluster 210, interconnecting module 220, be
System bus 230 and slave device 240 are all integrated in same chip.
When the processor 211 in processor cluster 210 will be written and read operation to Cache, 210 in processor cluster
Processor 211 will be sent out read-write requests, wherein, the read-write requests contain the mark of processor 211.Processor
211 read-write requests sent will relayed module 220 received.
120:With the foundation that is identified as of processor, query configuration table is to obtain virtual machine that processor is currently running
Mark, wherein, the mark for the virtual machine that the allocation list contains the mark of processor and processor is currently running is reflected
Penetrate relation.
Specifically, interconnecting module 220 is with the foundation that is identified as of processor 211, and query configuration table 222 is to obtain processor
The mark of 211 virtual machines being currently running.Asked for example, the processor CPU0 in processor cluster Cluster0 have issued read-write
Ask, then interconnecting module 220 removes query configuration table 0 according to the mark of the processor CPU0 in processor cluster Cluster0, and obtains
The virtual identifying VMID0 of the virtual machine run in processor CPU0 in processor cluster Cluster0.
130:The mark of virtual machine is sent to slave device so that the virtual machine that slave device is currently running according to processor
Mark to read-write operation carry out performance optimization processing.
Specifically, the virtual machine for obtaining running in the processor CPU0 in processor cluster Cluster0 in inquiry it is virtual
After identifying VMID0, slave device 240 is sent to by system bus 230 so that slave device 240 is currently being transported according to processor
The mark of capable virtual machine carries out performance optimization processing to read-write operation.
It is understood that in other implementations, processor cluster 210, interconnecting module 220, system bus 230 with
And slave device 240 can not also be integrated in same chip, but mutual independent hardware device.
The quantity of processor in processor cluster can be 1 or multiple, and the application is not especially limited.
In above-mentioned scheme, after receiving the read-write requests that the processor in processor cluster is sent, according to processor
Mark query configuration table to obtain the mark for the virtual machine that processor is currently running, it is then, current further according to processor
The mark for the virtual machine being currently running optimizes read-write operation.By implementing the embodiment of the present invention, can be matched somebody with somebody according to inquiry
The mark that table knows the virtual machine for sending read-write requests is put, the read-write requests that virtual machine sends are known according to the mark of virtual machine
The characteristics of feature, the read-write requests then sent according to virtual machine, optimizes read-write operation.
Referring to Fig. 3, Fig. 3 is the flow chart of another embodiment of read-write operation performance optimization method of the present invention.This embodiment party
The read-write operation performance optimization method of formula includes:
310:Create allocation list.
Specifically, referring to Fig. 2, as shown in Figure 2 in a multi-processor system, 16 processors are divided into 4
210,4 processor clusters 210 of processor cluster are respectively processor cluster Cluster0~Cluster3.Each processor cluster 210 wraps
It is respectively processor CPU0~CPU3 containing 4 processors, 211,4 processors 211.In interconnecting module 220, for each processing
Device cluster 210 generates an allocation list 222, so, there are totally 4 allocation lists 222 of allocation list 0~3.Allocation list 222 is used for record pair
Answer the mark for the virtual machine that the processor 211 in processor cluster 210 is currently running.Interconnecting module 220 also passes through system bus 230
It is connected with multiple slave devices 240.Wherein, slave device 240 can be Cache etc..Processor cluster 210, interconnecting module 220, be
System bus 230 and slave device 240 are all integrated in same chip.
Interconnecting module 220 is currently running virtual by the processor 211 that operating system acquisition processor 211 is sent
The mark of machine.Then, the virtual machine that interconnecting module 220 establishes the mark of processor 211 and processor 211 is currently running
Mapping relations between mark.Finally, the mark for the virtual machine mark of processor 211 and processor 211 being currently running
Mapping relations between knowledge are loaded into allocation list 222.
It is understood that interconnecting module 220, which can be each processor cluster 210, generates an allocation list 222, can also
An allocation list 222 is generated for multiple processor clusters 210, is not especially limited herein.
320:Receive the read-write requests that the processor in processor cluster is sent.
Specifically, when the processor 211 in processor cluster 210 will be written and read operation to Cache, positioned at processor cluster
Interior 210 processor 211 will be sent out read-write requests, wherein, the read-write requests contain the mark of processor 211.
The read-write requests that processor 211 is sent will relayed module 220 received.
330:With the foundation that is identified as of processor, query configuration table is to obtain virtual machine that processor is currently running
Mark, wherein, the mark for the virtual machine that the allocation list contains the mark of processor and processor is currently running is reflected
Penetrate relation.
Specifically, interconnecting module 220 is with the foundation that is identified as of processor 211, and query configuration table 222 is to obtain processor
The mark of 211 virtual machines being currently running.Asked for example, the processor CPU0 in processor cluster Cluster0 have issued read-write
Ask, then interconnecting module 220 removes query configuration table 0 according to the mark of the processor CPU0 in processor cluster Cluster0, and obtains
The virtual identifying VMID0 of the virtual machine run in processor CPU0 in processor cluster Cluster0.
340:The mark of virtual machine is sent to slave device so that the virtual machine that slave device is currently running according to processor
Mark to read-write operation carry out performance optimization processing.
Specifically, the virtual machine for obtaining running in the processor CPU0 in processor cluster Cluster0 in inquiry it is virtual
After identifying VMID0, slave device 240 is sent to by system bus 230 so that slave device 240 is current according to processor 211
The mark of the virtual machine of operation carries out performance optimization processing to read-write operation.Slave device 240 is currently being transported according to processor 211
The mark of capable virtual machine to read-write operation carry out performance optimization processing mode have many kinds, for example, (1) by slave device according to
The mark of virtual machine obtains virtual machine and is written and read operation rule, and is receiving next virtual machine read-write requests according to rule
Before, accessed data are got into Cache in advance.For example, the virtual machine that virtual identifying is VMID0 is by idol when being written and read operation
Number rule is written and read block address operation, then slave device is written and read operation by even rule to block address 0,2,4,6, and can
Before next virtual machine read-write requests, the data for the address space that block address is 8 are got cache in advance in advance and are delayed
Device is rushed, so as to improve the speed of read-write operation.(2) if operation scans class method in virtual machine, since scanning class method exists
For the first time to wherein some data block is written and read operation after, operation seldom is written and read to this data block again, so, limit
Virtual machine distribution cache buffer is made as, is used with saving memory headroom to other virtual machines.(3) by slave device root
Virtual machine distribution road (way) is identified as according to virtual machine, wherein, way is a structural units of cache buffer,
At least part virtual machine is assigned to different way.In the case of enough in space, each virtual machine can be made individually to take one
A or multiple way.But in the case of insufficient space, multiple virtual machines can also be made to share a way.It is for example, virtual
Machine 1 and virtual machine 2 share way1, and virtual machine 3 and virtual machine 4 share way2.Alternatively, virtual machine 1 monopolizes way1,1 He of virtual machine
Virtual machine 2 shares way2, and virtual machine 2 monopolizes way3 etc..
It is understood that the quantity of the processor in processor cluster can be 1 or multiple, the application is not
Make specific limit.
Referring to Fig. 4, Fig. 4 is the flow chart of the another embodiment of read-write operation performance optimization method of the present invention.This embodiment party
The read-write operation performance optimization method of formula includes:
410:Receive the mark for the virtual machine that the processor transmitted by interconnecting module is currently running.Wherein, virtual machine
Mark switching model allocation list in the block be stored in by interconnecting module inquiry obtain, allocation list contains the mark of the processor
The mapping relations of the mark for the virtual machine being currently running with the processor.
Specifically, by Fig. 1 and Fig. 3 shown in embodiment, interconnecting module obtains the void that processor is currently running
The mark of plan machine, and it is sent to slave device.What the processor transmitted by the correspondingly received interconnecting module of slave device was currently running
The mark of virtual machine.
420:The mark for the virtual machine being currently running according to processor carries out performance optimization processing.
Specifically, it is excellent to carry out performance to read-write operation for the mark for the virtual machine that slave device is currently running according to processor
The mode for changing processing has many kinds, for example, (1) obtains virtual machine by slave device according to the mark of virtual machine is written and read operation rule
Rule, and data are got by Cache according to rule in advance before next virtual machine read-write requests are received.For example, virtual identifying
To be written and read operation to block address by even rule when being written and read operation for the virtual machine of VMID0, then slave device presses even number
Rule is written and read block address 0,2,4,6 operation, and can be by block address in advance before next virtual machine read-write requests
The data of 8 address space get cache buffer in advance, so as to improve the speed of read-write operation.(2) it is if virtual
Operation scanning class method in machine, after for the first time to wherein some data block is written and read operation, seldom again to this data
Block is written and read operation, so, virtual machine distribution cache buffer is limited to, to save memory headroom to other empty
Plan machine uses.(3) virtual machine distribution road (way) is identified as according to virtual machine by slave device, wherein, way is slow at a high speed
A structural units of buffer is deposited, at least part virtual machine is assigned to different way., can in the case of enough in space
To make each virtual machine individually take one or more way.But in the case of insufficient space, multiple void can also be made
Plan machine shares a way.For example, virtual machine 1 and virtual machine 2 share way1, virtual machine 3 and virtual machine 4 share way2.Alternatively,
Virtual machine 1 monopolizes way1, and virtual machine 1 and virtual machine 2 share way2, and virtual machine 2 monopolizes way3 etc..
In above-mentioned scheme, after receiving the read-write requests that the processor in processor cluster is sent, according to processor
Mark query configuration table to obtain the mark for the virtual machine that processor is currently running, it is then, current further according to processor
The mark for the virtual machine being currently running optimizes read-write operation.By implementing the embodiment of the present invention, can be matched somebody with somebody according to inquiry
The mark that table knows the virtual machine for sending read-write requests is put, the read-write requests that virtual machine sends are known according to the mark of virtual machine
The characteristics of feature, the read-write requests then sent according to virtual machine, optimizes read-write operation.
The above-mentioned method for illustrating the embodiment of the present invention, following for easy to preferably implementing the embodiment of the present invention
Such scheme, correspondingly, is also provided below the relevant device for implementing such scheme for coordinating.
Referring to Fig. 5, Fig. 5 is the structure diagram of read-write operation performance optimization one embodiment of device of the present invention.This reality
Applying the read-write operation performance optimization device 500 of mode includes:Receiving module 510, enquiry module 520 and sending module 530.
The receiving module 510 is used to receive the read-write requests that the processor in processor cluster is sent, wherein, it is described
Read-write requests contain the mark of the processor, and the mark of the processor is sent to described look into by the receiving module 510
Ask module 520.
The enquiry module 520 is used for the mark for receiving the processor, with the foundation that is identified as of the processor, inquiry
Allocation list to obtain the mark for the virtual machine that the processor is currently running, wherein, the allocation list contains the place
The mapping relations of the mark for the virtual machine that the mark of reason device is currently running with the processor, the enquiry module 520 are used
In the mark of the virtual machine is sent to the sending module 530.
The sending module 530 is used for the mark for receiving the virtual machine, and the mark of the virtual machine is sent to slave device,
So that the mark for the virtual machine that the slave device is currently running according to the processor carries out performance optimization to read-write operation
Processing.
Device shown in Fig. 5 can perform each step of the method shown in Fig. 1, referring specifically to Fig. 1 and associated description,
Not reinflated description herein.
Refering to Fig. 6, read-write operation performance optimization device 600 further includes:Acquisition module 540, establish module 550 and loading mould
Block 560.
The acquisition module 540 is used to obtain the processor of processor transmission currently by operating system
The mark of the mark of the processor and the virtual machine is sent to by the mark of the virtual machine of operation, the acquisition module 540
It is described to establish module 550.
The module 550 of establishing is used to receive the mark of the processor and the mark of the virtual machine, establishes the place
Mapping relations between the mark for the virtual machine that the mark of reason device and the processor are currently running, it is described to establish module
The mapping relations are sent to the load-on module 560 by 550.
The load-on module 560 is used to receive the mapping relations, and the mark of the processor and the processor are worked as
Before mapping relations between the mark of virtual machine that is currently running be loaded into allocation list.
Alternatively, the virtual machine is obtained to cache buffer according to the mark of the virtual machine by the slave device
The rule of operation is written and read, and will be accessed in advance before next virtual machine read-write requests are received according to the rule
Data get cache buffer.
Alternatively, if operation scans class method in the virtual machine, the virtual machine is limited to by slave device and is distributed
Cache buffer.
Alternatively, road is distributed according to the virtual machine that is identified as of the virtual machine by slave device, wherein, the Lu Weigao
One structural units of fast cache buffer, at least part virtual machine are respectively allocated the different roads.
Device shown in Fig. 6 can perform each step of the method shown in Fig. 1 and Fig. 3, referring specifically to Fig. 1, Fig. 3 with
And associated description, not reinflated description herein.
In above-mentioned scheme, after receiving the read-write requests that the processor in processor cluster is sent, according to processor
Mark query configuration table to obtain the mark for the virtual machine that processor is currently running, it is then, current further according to processor
The mark for the virtual machine being currently running optimizes read-write operation.By implementing the embodiment of the present invention, can be matched somebody with somebody according to inquiry
The mark that table knows the virtual machine for sending read-write requests is put, the read-write requests that virtual machine sends are known according to the mark of virtual machine
The characteristics of feature, the read-write requests then sent according to virtual machine, optimizes read-write operation.
Referring to Fig. 7, Fig. 7 is the structure diagram of read-write operation performance optimization one embodiment of device of the present invention.This reality
Applying the read-write operation performance optimization device 700 of mode includes:Receiving module 710 and optimization module 720.
The receiving module 710 is used to receive the virtual machine that the processor transmitted by interconnecting module is currently running
Mark, wherein, the mark of the virtual machine is stored in switching model allocation list in the block by interconnecting module inquiry and obtains
, the mark for the virtual machine that the mark that the allocation list contains the processor is currently running with the processor is reflected
Relation is penetrated, the mark of the virtual machine is sent to the optimization module 720 by the receiving module 710.
The mark progress performance that the optimization module 720 is used for the virtual machine being currently running according to the processor is excellent
Change is handled.
Alternatively, the optimization module 720 is additionally operable to obtain the virtual machine at a high speed according to the mark of the virtual machine
Cache buffer is written and read the rule of operation, and according to the rule before next virtual machine read-write requests are received, in advance
Data are first got into cache buffer.
Alternatively, when the optimization module 720 is additionally operable to the operation scanning class method in the virtual machine, it is limited to described
Virtual machine distributes cache buffer.
Alternatively, the optimization module 720 is additionally operable to be identified as the virtual machine distribution road according to the virtual machine, its
In, the road be cache buffer a structural units, at least part virtual machine be respectively allocated it is different described in
Road.
Device shown in Fig. 7 can perform each step of the method shown in Fig. 4, referring specifically to Fig. 4 and associated description,
Not reinflated description herein.
In above-mentioned scheme, after receiving the read-write requests that the processor in processor cluster is sent, according to processor
Mark query configuration table to obtain the mark for the virtual machine that processor is currently running, it is then, current further according to processor
The mark for the virtual machine being currently running optimizes read-write operation.By implementing the embodiment of the present invention, can be matched somebody with somebody according to inquiry
The mark that table knows the virtual machine for sending read-write requests is put, the read-write requests that virtual machine sends are known according to the mark of virtual machine
The characteristics of feature, the read-write requests then sent according to virtual machine, optimizes read-write operation.
One of ordinary skill in the art will appreciate that realize all or part of flow in above-described embodiment method, being can be with
Relevant hardware is instructed to complete by computer program, the program can be stored in a computer read/write memory medium
In, the program is upon execution, it may include such as the flow of the embodiment of above-mentioned each method.Wherein, the storage medium can be magnetic
Dish, CD, read-only memory (Read-Only Memory, ROM) or random access memory (Random Access
Memory, RAM) etc..
The above disclosed power for being only a kind of preferred embodiment of the present invention, the present invention cannot being limited with this certainly
Sharp scope, one of ordinary skill in the art will appreciate that realizing all or part of flow of above-described embodiment, and is weighed according to the present invention
Profit requires made equivalent variations, still falls within and invents covered scope.
Claims (6)
- A kind of 1. read-write operation performance optimization method, it is characterised in that the described method includes:The read-write requests that the processor in processor cluster is sent are received, wherein, the read-write requests contain the processing The mark of device;With the foundation that is identified as of the processor, query configuration table is to obtain virtual machine that the processor is currently running Mark, wherein, the virtual machine that the mark that the allocation list contains the processor is currently running with the processor The mapping relations of mark;The mark of the virtual machine is sent to slave device so that the slave device obtains the void according to the mark of the virtual machine Plan machine is written and read cache buffer the rule of operation, and is receiving next virtual machine read-write according to the rule Before request, accessed data are got into cache buffer in advance;If operation scanning class method, limits in the virtual machine It is made as the virtual machine distribution cache buffer;The virtual machine distribution road is identified as according to the virtual machine, wherein, The road is a structural units of cache buffer, and at least part virtual machine is respectively allocated the different roads.
- 2. according to the method described in claim 1, it is characterized in that, receiving the reading of the transmission of the processor in processor cluster Before write request, the method further includes:The mark for the virtual machine that the processor that the processor sends is currently running is obtained by operating system;The mapping relations that the mark of the processor is established between the mark of virtual machine that the processor is currently running;Mapping relations between the mark for the virtual machine that the mark of the processor and the processor are currently running add It is downloaded in allocation list.
- A kind of 3. read-write operation performance optimization method, it is characterised in that the described method includes:The mark for the virtual machine that the processor transmitted by interconnecting module is currently running is received, wherein, the mark of the virtual machine Know and switching model allocation list acquisition in the block is stored in by interconnecting module inquiry, the allocation list contains the processing The mapping relations of the mark for the virtual machine that mark and the processor of device are currently running;The rule of operation, and root are written and read cache buffer according to the mark acquisition virtual machine of the virtual machine According to the rule before next virtual machine read-write requests are received, accessed data are got into cache buffer in advance;If operation scanning class method in the virtual machine, is limited to the virtual machine distribution cache buffer;The virtual machine distribution road is identified as according to the virtual machine, wherein, the road is one of cache buffer Structural units, at least part virtual machine are respectively allocated the different roads.
- 4. a kind of read-write operation performance optimizes device, it is characterised in that described device includes:Receiving module, enquiry module and Sending module:The receiving module is used to receive the read-write requests that the processor in processor cluster is sent, wherein, the read-write please The mark for containing the processor is sought, the mark of the processor is sent to the enquiry module by the receiving module;The enquiry module is used for the mark for receiving the processor, with the foundation that is identified as of the processor, query configuration table To obtain the mark for the virtual machine that the processor is currently running, wherein, the allocation list contains the processor The mapping relations of the mark for the virtual machine that mark is currently running with the processor, the enquiry module are used for the void The mark of plan machine is sent to the sending module;The sending module is used for the mark for receiving the virtual machine, the mark of the virtual machine is sent to slave device so that institute The rule that slave device is written and read cache buffer according to the mark acquisition virtual machine of the virtual machine operation is stated, And according to the rule before next virtual machine read-write requests are received, accessed data are got into cache buffering in advance Device;If operation scanning class method in the virtual machine, is limited to the virtual machine distribution cache buffer;According to institute That states virtual machine is identified as the virtual machine distribution road, wherein, the road is a structural units of cache buffer, extremely Small part virtual machine is respectively allocated the different roads.
- 5. device according to claim 4, it is characterised in that described device further includes:Acquisition module, establish module and add Carry module:What the acquisition module was currently running for obtaining the processor of the processor transmission by operating system The mark of the mark of the processor and the virtual machine is sent to and described establishes mould by the mark of virtual machine, the acquisition module Block;The module of establishing is used to receive the mark of the processor and the mark of the virtual machine, establishes the mark of the processor Mapping relations between the mark for the virtual machine that knowledge and the processor are currently running, the module of establishing is by the mapping Relation is sent to the load-on module;The load-on module is used to receive the mapping relations, and the mark of the processor and the processor are currently being transported Mapping relations between the mark of capable virtual machine are loaded into allocation list.
- 6. a kind of read-write operation performance optimizes device, it is characterised in that described device includes:Receiving module and optimization module:The receiving module is used for the mark for receiving the virtual machine that the processor transmitted by interconnecting module is currently running, its In, the mark of the virtual machine is stored in switching model allocation list in the block by interconnecting module inquiry and obtains, described to match somebody with somebody Put the mapping relations that table contains the mark for identifying the virtual machine being currently running with the processor of the processor, institute State receiving module and the mark of the virtual machine is sent to the optimization module;The optimization module is additionally operable to obtain the virtual machine to cache buffer progress according to the mark of the virtual machine The rule of read-write operation, and according to the rule before next virtual machine read-write requests are received, in advance by accessed data Get cache buffer;In the virtual machine during operation scanning class method, it is slow at a high speed to be limited to the virtual machine distribution Deposit buffer;The virtual machine distribution road is identified as according to the virtual machine, wherein, the road is cache buffer One structural units, at least part virtual machine are respectively allocated the different roads.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201410419211.9A CN104216758B (en) | 2014-08-22 | 2014-08-22 | A kind of read-write operation performance optimization method and device |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201410419211.9A CN104216758B (en) | 2014-08-22 | 2014-08-22 | A kind of read-write operation performance optimization method and device |
Publications (2)
Publication Number | Publication Date |
---|---|
CN104216758A CN104216758A (en) | 2014-12-17 |
CN104216758B true CN104216758B (en) | 2018-05-04 |
Family
ID=52098286
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201410419211.9A Active CN104216758B (en) | 2014-08-22 | 2014-08-22 | A kind of read-write operation performance optimization method and device |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN104216758B (en) |
Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN101561769A (en) * | 2009-05-25 | 2009-10-21 | 北京航空航天大学 | Process migration tracking method based on multi-core platform virtual machine |
CN102483718A (en) * | 2009-08-25 | 2012-05-30 | 国际商业机器公司 | Cache partitioning in virtualized environments |
CN103902352A (en) * | 2012-12-28 | 2014-07-02 | 华为技术有限公司 | Method and device for acquiring request data in multi-core virtual machine system |
Family Cites Families (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20090077550A1 (en) * | 2007-09-13 | 2009-03-19 | Scott Rhine | Virtual machine schedular with memory access control |
-
2014
- 2014-08-22 CN CN201410419211.9A patent/CN104216758B/en active Active
Patent Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN101561769A (en) * | 2009-05-25 | 2009-10-21 | 北京航空航天大学 | Process migration tracking method based on multi-core platform virtual machine |
CN102483718A (en) * | 2009-08-25 | 2012-05-30 | 国际商业机器公司 | Cache partitioning in virtualized environments |
CN103902352A (en) * | 2012-12-28 | 2014-07-02 | 华为技术有限公司 | Method and device for acquiring request data in multi-core virtual machine system |
Also Published As
Publication number | Publication date |
---|---|
CN104216758A (en) | 2014-12-17 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
US11349940B2 (en) | Server side data cache system | |
US20200358710A1 (en) | Container-based mobile code offloading support system in cloud environment and offloading method thereof | |
CN104461958B (en) | Support storage resource access method, storage control and the storage device of SR-IOV | |
US8943294B2 (en) | Software architecture for service of collective memory and method for providing service of collective memory using the same | |
CN104871143B (en) | System and method for distributing storage to dissimilar storage devices using service quality | |
CN102255962B (en) | Distributive storage method, device and system | |
KR20210147976A (en) | Systems and methods for scalable and coherent memory devices | |
US7958329B2 (en) | Hybrid replicated shared memory | |
US20150127691A1 (en) | Efficient implementations for mapreduce systems | |
CN103049422A (en) | Method for building multi-processor node system with multiple cache consistency domains | |
CN109542814A (en) | The method and system that data are transmitted between storage equipment is connected to by the P2P of PCI-EXPRESS | |
CN106648462B (en) | Date storage method and device | |
CN108509152A (en) | A kind of method and device for realizing the storage of container runtime data | |
CN108319496A (en) | resource access method, service server, distributed system and storage medium | |
CN105335170A (en) | Distributed system and incremental data updating method | |
CN102063386B (en) | Cache management method of single-carrier multi-target cache system | |
CN103297490B (en) | Information processing apparatus, distributed processing system, and distributed processing method | |
US20190272238A1 (en) | Determining memory access categories for tasks coded in a computer program | |
CN103488523A (en) | Page access method, page access device and server | |
CN106960011A (en) | Metadata of distributed type file system management system and method | |
WO2020214757A1 (en) | Virtual memory pool within a network which is accessible from multiple platforms | |
US20180074851A1 (en) | Determining memory access categories to use to assign tasks to processor cores to execute | |
WO2023200502A1 (en) | Direct swap caching with zero line optimizations | |
US11061676B2 (en) | Scatter gather using key-value store | |
US20100121935A1 (en) | Hybrid replicated shared memory |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
C06 | Publication | ||
PB01 | Publication | ||
C10 | Entry into substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
GR01 | Patent grant | ||
GR01 | Patent grant |