CN105516223B - Virtual storage system and its implementation, server and monitor of virtual machine - Google Patents
Virtual storage system and its implementation, server and monitor of virtual machine Download PDFInfo
- Publication number
- CN105516223B CN105516223B CN201410497434.7A CN201410497434A CN105516223B CN 105516223 B CN105516223 B CN 105516223B CN 201410497434 A CN201410497434 A CN 201410497434A CN 105516223 B CN105516223 B CN 105516223B
- Authority
- CN
- China
- Prior art keywords
- bootvdi
- client
- data
- cache blocks
- server
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
Landscapes
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
- Information Transfer Between Computers (AREA)
- Stored Programmes (AREA)
Abstract
The present invention discloses a kind of virtual storage system and its implementation, server and monitor of virtual machine.The virtual storage system implementation method includes: to be buffered in BootVDI data in a manner of cache blocks in the cooperative caching of monitor of virtual machine from server download system virtual disk image BootVDI data；The BootVDI read request from client operating system is received, includes the caching block identification of BootVDI data in BootVDI read request；The cache blocks of the BootVDI data cached in cooperative caching are supplied to client operating system.By the way that the cooperative caching for saving system virtualization disk image BootVDI data is arranged in the monitor of virtual machine of each client, and shared cooperative caching, each client is enabled to read the BootVDI of other clients as needed, the amount of access to server is reduced, the reading efficiency of virtual storage system is optimized.
Description
Technical field
The present invention relates to field of cloud calculation more particularly to a kind of virtual storage systems and its implementation, server and void
Quasi- monitor unit.
Background technique
In cloud computation data center, the virtual memory space in client is mapped to the virtual disk image on server
(Virtual Disk Image, VDI), the read-write requests to VDI are the main sources of server load.Depending on the application not
Together, VDI can be divided into two class of virtual disk image BootVDI and user's virtual disk UserVDI, wherein BootVDI is responsible for preservation
Operating system and program, os starting and program data required when loading are obtained by reading BootVDI mostly.
Since server bandwidth and processing capacity are limited, when multiple client carries out the operations such as os starting, program load simultaneously
When, the contention of server resource will cause a large amount of requests and be lined up in the buffer, cause the response time to be significantly increased, so that reading efficiency
It is low, influence the performance of virtual storage system.
Therefore, it is necessary to propose a kind of method to optimize the reading efficiency of BootVDI, access of the client to server is reduced
Amount, improves the performance of virtual storage system.
Summary of the invention
The disclosure technical problem to be solved is how to optimize the reading efficiency of BootVDI, reduces the visit to server
The amount of asking improves the performance of virtual storage system.
The disclosure provides a kind of virtual storage system implementation method, comprising:
From server download system virtual disk image BootVDI data, BootVDI data are delayed in a manner of cache blocks
There are in the cooperative caching of monitor of virtual machine；
The BootVDI read request from client operating system is received, includes BootVDI data in BootVDI read request
Caching block identification；
The cache blocks of the BootVDI data cached in cooperative caching are supplied to client operating system.
In one embodiment, which can also include:
History temperature of the server according to cache blocks, the BootVDI data block that each client of equilibrium assignment need to cache, and will
Allocation result generates global buffer and accesses table GCT；
The BootVDI data block and GCT that each client need to cache are sent to each client by server.
In one embodiment, server need to be cached according to the history temperatures of cache blocks, each client of equilibrium assignment
BootVDI data block, comprising:
It is equal to distribute DPSB algorithm by dynamic equilibrium temperature and spatial cache block according to the history temperature of cache blocks for server
Weighing apparatus distributes the BootVDI data block that each client need to cache；
In one embodiment, further includes: server needs to cache according to the history temperatures of cache blocks, each client of equilibrium assignment
BootVDI data block, comprising: server is more than the cache blocks point of first threshold according to the history temperatures of cache blocks, by temperature
Solution is at least two cache blocks copies, and makes cache blocks copy and cache blocks address having the same.
In one embodiment, which includes:
The BootVDI read request of client operating system is received, GCT is inquired, determines the ground of the cache blocks of BootVDI data
Location；
Judge whether the cache blocks for caching BootVDI data overload, and judges whether to hit BootVDI data；
If caching the cache blocks overload or miss BootVDI data of BootVDI data, BootVDI forwarding request is sent
To server.
In one embodiment, server receives BootVDI forwarding request, forwards request to return to BootVDI according to BootVDI
Data and control command, control command include cache blocks replacement order and update GCT order.
In one embodiment, further includes: receive BootVDI data and control command that server returns, execute control life
Enable, by BootVDI data buffer storage in the cache blocks of cooperative caching, replacement temperature be less than second threshold cache blocks copy or
Only existing copy, and update GCT table.
The disclosure also provides a kind of monitor of virtual machine, comprising:
Network unit, for from server download system virtual disk image BootVDI data, by BootVDI data with
The mode of cache blocks is buffered in the cooperative caching of monitor of virtual machine；
Read-write cell is connected with network unit, for receiving the BootVDI read request from client operating system,
It include the caching block identification of BootVDI data in BootVDI read request；
Processing unit is connected with read-write cell and network unit, the BootVDI number for will cache in cooperative caching
According to cache blocks be supplied to client operating system.
Optionally, network unit is also used to:
The temperature of cache blocks is sent to server；
Receive the BootVDI data and global buffer access table GCT that server is sent, wherein server is according to cache blocks
History temperature, the BootVDI data block that each client of equilibrium assignment need to cache, and allocation result is generated into GCT, by each client
The BootVDI data block that need to be cached and GCT is held to be sent to each client.
Optionally, processing unit is also used to the history temperature according to each cache blocks, is more than the caching of first threshold by temperature
Block is decomposed at least two cache blocks copies, and makes cache blocks copy and cache blocks address having the same.
Optionally, processing unit is also used to:
According to the BootVDI read request of the client operating system received, GCT is inquired, determines the slow of BootVDI data
The address of counterfoil；
Judge whether the cache blocks for caching BootVDI data overload, and judges whether to hit BootVDI data；
If caching the cache blocks overload or miss BootVDI data of BootVDI data, BootVDI forwarding request is sent
To server.
Optionally, network unit is also used to send BootVDI forwarding request to server, wherein if server receives
BootVDI forwarding request forwards request to return to BootVDI data and control command according to BootVDI, and control command includes caching
Block replacement order and update GCT order.
Optionally, processing unit is also used to: being received BootVDI data block and control command that server returns, is executed control
System order, by BootVDI data buffer storage in the cache blocks of cooperative caching, replacement temperature is less than the cache blocks pair of second threshold
Sheet or only existing copy, and update GCT table.
The disclosure also provides a kind of virtual storage system server characterized by comprising
Input-output unit, for sending system virtualization disk image BootVDI data, by BootVDI data to cache
The mode of block is stored in the cooperative caching of the monitor of virtual machine of each client；
Administrative unit is connected with input-output unit, for the history temperature according to cache blocks, each client of equilibrium assignment
Hold the BootVDI data block that need to be cached.
Optionally, administrative unit is also used to the history temperature according to cache blocks, passes through dynamic equilibrium temperature and spatial cache
The BootVDI data block that block distribution each client of DPSB algorithm equilibrium assignment need to cache, and allocation result is generated into global buffer
Access table GCT；
Input-output unit is also used to BootVDI data block and GCT that each client need to cache being sent to each client.
Optionally, administrative unit is also used to the history temperature according to cache blocks, is more than the cache blocks of first threshold by temperature
At least two cache blocks copies are decomposed into, and make cache blocks copy and cache blocks address having the same.
Optionally, administrative unit is used for:
If receiving the BootVDI read request of client operating system, BootVDI number is returned to according to BootVDI read request
According to and control command, control command include cache blocks replacement order and update GCT order.
Optionally, administrative unit is used for:
If receiving BootVDI forwarding request, request is forwarded to return to BootVDI data and control command according to BootVDI,
Control command includes cache blocks replacement order and update GCT order.
Optionally, input-output unit is also used to send BootVDI data and control command, client executing to client
Control command, by BootVDI data-block cache in the cache blocks in cooperative caching, replacement temperature is less than the slow of second threshold
Counterfoil copy or only existing copy, and update GCT table.
The disclosure also provides a kind of virtual storage system, comprising: above-mentioned monitor of virtual machine；Above-mentioned virtual memory system
System server.
The virtual storage system and its implementation, server and monitor of virtual machine that the disclosure provides, by each visitor
Cooperative caching for saving system virtualization disk image BootVDI data is set in the monitor of virtual machine at family end, and altogether
Cooperative caching is enjoyed, each client is enabled to read the BootVDI of other clients as needed, reduces the access to server
Amount, optimizes the reading efficiency of virtual storage system.
Detailed description of the invention
Fig. 1 shows the flow chart of the virtual storage system implementation method of one embodiment of the invention.
Fig. 2 shows the structural schematic diagrams of the virtual storage system of one embodiment of the invention.
Fig. 3 shows the structural schematic diagram of the virtual storage system of one embodiment of the invention.
Fig. 4 shows the flow chart of the virtual storage system implementation method of another embodiment of the present invention.
Fig. 5 shows the DPSB algorithm of the virtual storage system implementation method of one embodiment of the invention.
Fig. 6 shows the flow chart of the virtual storage system implementation method of another embodiment of the present invention.
Fig. 7 shows the flow chart of the virtual storage system implementation method of another embodiment of the present invention.
Fig. 8 shows the structural block diagram of the virtual storage system client of one embodiment of the invention.And
Fig. 9 shows the structural block diagram of the virtual storage system server of one embodiment of the invention.
Specific embodiment
With reference to the accompanying drawings to invention is more fully described, wherein illustrating exemplary embodiment of the present invention.
Fig. 1 shows the flow chart of the virtual storage system implementation method of one embodiment of the invention.As shown in Figure 1, this method
It specifically includes that
Step S100, from server download system virtual disk image BootVDI data, by BootVDI data to cache
The mode of block is buffered in the cooperative caching of monitor of virtual machine.
Step S102, the BootVDI read request from client operating system is received, includes in BootVDI read request
The caching block identification of BootVDI data.
Step S104, the cache blocks of the BootVDI data cached in cooperative caching are supplied to client operating system.
Specifically, cooperative caching is set at monitor of virtual machine (Virtual Machine Monitor, VMM)
(CoCache, alternatively referred to as global buffer), for save system virtualization disk image (Boot Virtual Disk Image,
BootVDI) data, using this cooperative caching mechanism, to optimize the read operation for sharing memory space in cloud computing.Wherein,
Cooperative caching strategy is the memory cache in the shared VMM of each client, shares the client with same requirements
The BootVDI data block downloaded to optimizes BootVDI read request to reduce the amount of access to server.Wherein,
BootVDI data block is the block for storing BootVDI data, is used to store the data of BootVDI data in cooperative caching
Block is properly termed as cache blocks.
It should be noted that client operating system can be local client operating system, it is also possible to other clients
The operating system at end.Local client operating system can also have multiple, such as windows, linux and mac os operation system
System.
Fig. 2 shows the structural schematic diagram of the virtual storage system of one embodiment of the invention, component identical with Fig. 1 in Fig. 2
Or step function having the same omits the detailed description to these components or step for simplicity.As shown in Fig. 2,
The virtual storage system includes at least one server and multiple client such as customer end A, customer end B and client C.In each visitor
A certain size memory can be set in the monitor of virtual machine VMM at family end as cooperative caching, cooperative caching is for saving
BootVDI data.For example, a memory cache is respectively arranged in the VMM of each client such as customer end A, customer end B and client C
Shared state is set as cooperative caching, and by the cooperative caching.Multiple client and server are connected by network
It connects, local client or other clients can be accessed freely mutually.In this way, client can read other clients
On BootVDI data, to avoid access server.
In one embodiment, when customer end A, customer end B and client C carry out the behaviour such as os starting, program load
When making, from server download system virtual disk image BootVDI data, BootVDI data are stored in a manner of cache blocks
In the cooperative caching of VMM.When the VMM of customer end A receives BootVDI read request, according to the caching of BootVDI data
Block identification (such as using buffer address as mark or cashing indication code as mark), preferentially local cooperative caching in look into
Data corresponding to BootVDI read request are ask, if not inquiring data corresponding to BootVDI read request, basis locally
Setting rule is inquired in customer end B, C, such as can be determined to send to customer end B or C according to the busy or spare time of client and be somebody's turn to do
BootVDI read request.If customer end B has BootVDI data corresponding to the BootVDI read request, customer end B should
BootVDI data are sent to customer end A, complete reading process；If not finding the BootVDI data in customer end B, can incite somebody to action
Read request is sent to client C to search above-mentioned BootVDI data, and successively poll preserves each client of the BootVDI data
End.If not inquiring BootVDI data corresponding to BootVDI read request in each client, sent out by the BootVDI read request
It send to server, server returns to BootVDI data according to the BootVDI read request.
The virtual storage system implementation method of the present embodiment, is used for by being arranged in the monitor of virtual machine of each client
The cooperative caching of system virtualization disk image BootVDI data, and shared cooperative caching are saved, enables each client root
According to the BootVDI for needing to read other clients, the amount of access to server is reduced, the reading efficiency of virtual storage system is optimized.
Fig. 3 shows the structural schematic diagram of the virtual storage system of one embodiment of the invention, identical with Fig. 1 or 2 group in Fig. 3
Part or step function having the same omit the detailed description to these components or step for simplicity.
As shown in figure 3, server 317 may include static load balancing device 319, BootVDI cache blocks information database
320, cooperative caching manager 321, extension I/O service 322, virtual disk image 318, virtual disk image 318 etc., wherein
Virtual disk image 318 include BootVDI 323, user's virtual disk image (User Virtual Disk Image,
UserVDI)324.UserVDI 324 is the virtual disk image of user data, and BootVDI323 is system disk (boot disk)
Virtual disk image, os starting and program data required when loading are obtained by reading BootVDI 323 mostly
?.The operating system and application program run in different clients is similar, this just makes same memory block in BootVDI 323
It may while successively be downloaded by multiple client.
Extension I/O service 322 is extended on the basis of the I/O service on virtual storage system server to caching block access
The monitoring function of frequency, it by the BootVDI read request information received pass to cooperative caching manager 321 carry out processing and
It saves, and the data in server in BootVDI data block can be sent to each client.Cooperative caching manager 321
It is responsible for being monitored loaded work piece circumstances and the replacement and layout of cooperative caching, it can determine that a cache blocks are
No to reside in client cooperative caching, it safeguards BootVDI cache blocks database, has recorded nearest following period of time in library
The temperature of each cache blocks.It is flat to execute dynamic according to the information in BootVDI cache blocks database for static load balancing device 319
The temperature that weighs and spatial cache block distribution (Dynamic Popularity and Storage Balance algorithm, DPSB)
Algorithm is periodically generated allocation plan.
As shown in figure 3, client 300 includes the moulds such as client operating system 301, virtual disk 306 and VMM 307
Block.Module work relevant to cooperative caching is in the VMM 307 of client.Client 300 is both the requestor of data,
It is the supplier of data.
Client operating system 301 includes application program 302, file system 303, virtual memory management 304 and virtual
Disk interface 305 etc..VMM 307 includes read-write (input and output, IO) processor 308, end I/O service 309, request
Initiation module 314, cooperative caching 316, global buffer access 313, network module table (Global Cache Table, GCT)
315 etc..It reads and writes processor 308 to be responsible for receiving the read-write requests of client operating system 301, is classified to read-write requests, arranged
Team and scheduling.End I/O service 309 is mainly responsible for the cooperative caching managed in each client VMM 307, and can handle other
The request of client.Holding I/O service 309 may include overload protection module 310, load monitoring module 311 and cooperative caching
Management module 312.Wherein, overload protection module 310 is after receiving the request according to the loading condition of client, determine processing or
Forward the request；Load monitoring module 311 is responsible for monitoring input rank length, counts the read request of a period of time processing recently
Number, judges whether system overloads, and adjusts system mode；The reading that cooperative caching management module 312 is responsible for caching is held with replacement
Row, it checks whether requested caching hits according to local cache mapping table, and the ground of hit cache blocks in memory
Then location executes reading or replacement operation.It requests initiation module 314 to access table 313 by global buffer and positions destination client,
Then data needed for communicating acquisition with destination client.Network module 315 can be VMM 307 and be based on network store system
(Network File System, NFS) or small computer system interface (Internet Small Computer
System Interface, iSCSI) etc. systems module, can be responsible for realizing client 300 and server 317 or client
300 with the communication of other clients.
Cooperative caching 316 be used for cache from server 317 downloading BootVDI data, and by the BootVDI data with
The form of cache blocks is stored in cooperative caching 316.If VMM receives the BootVDI from client operating system, reading is asked
It asks, according to the cache blocks identifier lookup cooperative caching of BootVDI data, if being stored with this BootVDI pairs in cooperative caching
The cache blocks for being buffered in BootVDI data in cooperative caching are then supplied to client operating system by the data answered.GCT
313 for storing the cache blocks position of cooperative caching, and GCT 313 is the linear list query caching indexed by cache blocks serial number
Block position, each list item have recorded the client-side information for saving corresponding cache blocks.Server 317 is warm according to the history of cache blocks
Degree, the BootVDI data block that each client of equilibrium assignment need to cache, and allocation result is generated into global buffer and accesses table GCT；?
When client terminal start-up, the BootVDI data block that each client need to cache and GCT table are pushed to each client by server, storage
In the VMM of client.
In one embodiment, the VMM 307 of client 300 is by inquiry GCT 313 come the position of the cache blocks of search request
It sets.GCT 313 is the linear list query caching block position indexed by cache blocks serial number, and it is corresponding slow that each list item has recorded preservation
The client-side information of counterfoil, but when encountering a cache blocks when storing the cooperative caching in multiple client, source client
It can be not at the client of " overload " state with random request one, BootVDI read request is sent, if not looking into a client
The corresponding BootVDI data of the BootVDI read request are found, can be according to the busy extent of each client, successively poll is each
Client, until finding above-mentioned BootVDI data.GCT 313 is safeguarded that it starts in client 300 by server 317
In the process, client is pushed to by server 317.In operational process, server 317 is every to be will pass through extensively every T at regular intervals
It broadcasts and updates a GCT.Since renewal time interval may be longer, the GCT information in client 300 can be expired, this to receive
BootVDI read request message may at this moment be needed to forward without requested cache blocks in the client of BootVDI read request
To the processing of server 317.Server 317 by forward come request learn that member record is already expired in the middle part of the GCT of certain client 300
Phase, therefore the neighbouring GCT segment of accessed block is attached in processing result return source client, allow source client more
New GCT.In addition, server 317 monitors the cache location of all clients, the frequent degree that can be changed according to client is adjusted
The whole frequency for updating GCT.
Algorithm is replaced in cooperative caching mainly considers load balance and global hit rate, it is unified to each by server 317
The cache policy of client carries out decision.In the case where general normal use, after client downloads cache blocks, it can be not preserved in
In cooperative caching.If but meet following two situation, can be replaced operation: (1) cache blocks downloaded to are only existing copies, can
To replace the copy that temperature is minimum in same client, without replacement if without copy；(2) according to DPSB algorithm, work as generation
When " overload " forwards, the cache blocks in client with similar temperature are replaced with the cache blocks downloaded to.
Since the virtual storage system implementation method based on cooperative caching works in virtual machine system level, void is adapted to
The demand that quasi- machine stores block grade.Since what is run on virtual-machine client is end operating system and application program, in order to make
Program smooth can be run, it is necessary to assure and the response time of request cannot be too long, the cooperative caching in virtual storage system,
Global hit rate is given priority to, next is only partial hit rate, can be to avoid function caused by N-Chance GMS scheduling algorithm
Repetition and problem of resource waste.Also, virtual storage system works in block layer, and virtual machine can perceive file semantics, based on negative
The DPSB algorithm for carrying equilibrium principle, can solve existing other algorithms not can be used directly asking in virtual storage system
Topic.
It is slow that cooperative is arranged in the virtual storage system implementation method of the present embodiment in the monitor of virtual machine of client
It deposits, by the temperature of the balanced each client of load balancing strategy, so that different clients temperature is uniform as far as possible, each client energy
The BootVDI of other clients is read as needed, it is possible to reduce client optimizes to the amount of access of server and reads efficiency, mention
High virtual storage system performance.
Fig. 4 shows the flow chart of the virtual storage system implementation method of another embodiment of the present invention.It is identical as Fig. 3 in Fig. 4
Component or step function having the same omit detailed description to these components or step for simplicity.Such as figure
Shown in 4, this method is specifically included that
Step S400, history temperature of the server according to cache blocks, the BootVDI number that each client of equilibrium assignment need to cache
According to block, and allocation result is generated into global buffer and accesses table GCT.
Specifically, load balance, the as far as possible caching in guarantee different clients are fully considered when designing distributed caching
Block temperature summation is close, avoids the occurrence of thermal center point.The temperature of cache blocks is frequency accessed within a certain period of time, is represented slow
The popular degree of counterfoil, the temperature of client are the sum of all cache blocks temperatures in client.In cooperative caching, a visitor
The probability that family end is accessed by other clients, it is closely related with its temperature.Client temperature is kept to be evenly distributed, it can be advantageous
Load balance between guaranteeing client.The target of load balance is to calculate cache blocks according to the historical data of cache blocks temperature
The optimal case distributed between client.Optimal meaning includes two o'clock: the first, different clients temperature is uniform as far as possible；The second,
Cache blocks occupy on the client memory space in other words cache blocks number also as far as possible uniformly.
In one embodiment, a kind of DPSB algorithm is proposed, the temperature to guarantee different clients is uniform.Fig. 5 shows this hair
The DPSB algorithm of the virtual storage system implementation method of a bright embodiment.As shown in figure 5, the algorithm main thought is server root
According to the history temperature of cache blocks, distributing each client of DPSB algorithm equilibrium assignment by dynamic equilibrium temperature and spatial cache block is needed
The BootVDI data block of caching.By the cache blocks that temperature is more than first threshold (for example, read number be more than 100 times) be decomposed into
Few two cache blocks copies, and make cache blocks copy and cache blocks address having the same, and allocation result is generated into GCT.?
After client terminal start-up, GCT is distributed to client by server, client can search the position of cache blocks according to GCT.
5 to 12nd line code is the main body of DPSB algorithm, wherein the complexity of the 5th row sort algorithm is O (I*lgI), 7 are arrived
12 rows are three layers of nested circulations, and K element sequence of eighth row, complexity is O (K*lgK), so the complexity of 7 to 12 row algorithms
For O (I*N*K*lgK).Since N, K are much smaller than I, so the complexity of DPSB algorithm is O (I*lgI), it can be in polynomial time
It completes.
Step S402, the BootVDI data block and GCT that each client need to cache are sent to each client by server.
Specifically, in practical application, after client terminal start-up, server needs the number cached to client push according to GCT
According to block.Static cache blocks deployment is completed before client operating system starts.
Mechanism guarantees that client will not become " hot spot " when providing two operations in cooperative caching.The first, each client
End monitoring request arriving rate, specific targets are the length L and average queue length Lavg of request queue.When L is greater than maximum queue
Length Lmax or be greater than the average queue length Lavg upper limit when, client will enter " overload " state.Under overload
Client, after receiving BootVDI read request, without processing, but the BootVDI read request is transmitted to server.The
Two, server by forward come read request, learn that client is in overload, adjustable cache blocks distribution increases
Overload the number of copies of cache blocks in client.Specifically method, which can be to work as, receives the request that hot customer terminal forwards
When, server sends control command while returning to request data, and notice initiates the client (i.e. source client) of request
Retain requested cache blocks.
After source client receives the reply of server, retain requested cache blocks, and page displacedment is least commonly used
It is minimum that algorithm (least frequently used page-replacement algorithm, LFU) algorithm replaces access
Copy.The principle of LFU algorithm is the displacement the smallest cache blocks copy of reference count in the displacement of cache blocks copy, this is because
The cache blocks copy being commonly used should have a biggish reference number, and the cache blocks copy being not frequently used should have one
It is least slow can to replace access times by replacing the smallest cache blocks copy of reference count for a lesser reference number
Counterfoil copy is conducive to reasonably adjust cooperative caching memory space in this way, and efficiency is read in optimization.But some buffer stopper copies exist
It is many to start with number, but does not just use later, this kind of buffer stopper copy will stay for a long time in memory, for this
Reference count register timing can be moved to right one, form the average access times of exponential damping by the buffer stopper copy of sample.Such as
Buffer stopper copy is not present in fruit, then replaces only existing copy with same algorithm, and in certain time interval T, avoids access thermal center point.
It is slow that cooperative is arranged in the virtual storage system implementation method of the present embodiment in the monitor of virtual machine of client
It deposits, by the temperature of the balanced each client of load balancing strategy, so that different clients temperature is uniform as far as possible, each client energy
The BootVDI of other clients is read as needed, it is possible to reduce client optimizes to the amount of access of server and reads efficiency, mention
High virtual storage system performance；In addition, passing through the strategy of central server management and distributed lookup, guaranteeing the response time
Meanwhile the expense of server can be reduced to the greatest extent.
Fig. 6 shows the flow chart of the virtual storage system implementation method of another embodiment of the present invention.In Fig. 6 with Fig. 1 to 4 phases
Same component or step function having the same omits the detailed description to these components or step for simplicity.Such as
Shown in Fig. 6, this method is specifically included that
Step S600, the BootVDI read request of client operating system is received, GCT is inquired, determines BootVDI data
Address.
Step S602, judge whether the cache blocks for caching BootVDI data overload, and judge whether to hit BootVDI
Data.
If step S604, caching the cache blocks overload or miss BootVDI data of BootVDI data, BootVDI is sent
Forward requests to server.
Specifically, if receiving the BootVDI read request of client operating system, the distributed lookup in each client
The corresponding data of BootVDI read request are forwarded in the case where client overload or miss, then by BootVDI read request
Server.In this way, it is possible to reduce to the amount of access of server, efficiency is read in optimization, improves virtual storage system performance.
Step S606, server receives BootVDI forwarding request, forwards request to return to BootVDI data according to BootVDI
And control command, control command include cache blocks replacement order and update GCT order.
Step S608, client receives the BootVDI data and control command of server return, executes control command,
By BootVDI data buffer storage in the cache blocks of cooperative caching, replacement temperature is less than cache blocks copy or the orphan of second threshold
This, and update GCT table.
Specifically, when holding operating system to read request to up to VMM, the read-write processor of VMM is classified, if it is access
The read request of BootVDI then gives request initiation module processing.Request initiation module inquires GCT first, if requested caching
Block then gives end I/O service in local, completes request after obtaining data；If requested cache blocks in other clients,
The non-overloaded destination client of random selection one sends BootVDI read request.
After destination client receives the BootVDI read request of source client forwarding, whether judgement currently overloads, Yi Jiyao
Whether the cache blocks of request are in the cooperative caching of oneself.If current state overload or BootVDI read request are not ordered
In, then forward the request to server.Otherwise, destination client returns to request results and gives source client.Server receives
After BootVDI read request, reads BootVDI and obtain data in cache blocks, and the reason of according to sending, in returned packet
Add corresponding control command.These control commands include cache blocks replacement order, update GCT segment etc..Source client or mesh
After mark client receives the result of return, the corresponding command is executed, a request process is completed.
The virtual storage system implementation method of the present embodiment, central server are requested according to the BootVDI forwarding received,
The cooperative caching and GCT table of each client are managed respectively, each client can respond the control command of central server,
Execute distributed management operation, the method combined by central management and distributed lookup, enable each client according to
It needs to read the BootVDI of other clients, reduces the amount of access to server, efficiency is read in optimization, improves virtual storage system
Performance reduces the expense of server to the greatest extent while guaranteeing the response time.
Fig. 7 shows the flow chart of the virtual storage system implementation method of another embodiment of the present invention.It is identical as Fig. 6 in Fig. 7
Component or step function having the same omit detailed description to these components or step for simplicity.Such as figure
Shown in 7, this method is specifically included that
Step S701, source client receives BootVDI memory space read request.
Step S702, according to GCT table, inquire local cache and hit? if the corresponding data storage of BootVDI read request
In local cooperative caching, then directly read locally, and execute step S717；If miss thens follow the steps S703.
Step S703, according to GCT table, inquire cooperative caching and hit? if it is corresponding to find BootVDI read request
Data are stored in destination client, then follow the steps S704；If not finding the corresponding data of BootVDI read request, execute
The BootVDI read request is sent to server by step S710.
Step S704, BootVDI read request is sent to destination client.
Step S705, destination client receives BootVDI read request.
Step S706, whether destination client inquiry overloads.If so, thening follow the steps S709, BootVDI read request is turned
It is sent to server；Otherwise, step S707 is executed.
Step S707, the machine cooperative caching is inquired in destination client, it is determined whether it is corresponding to hit the BootVDI read request
Data, if hit, then follow the steps S708；Otherwise, it executes step S709 and BootVDI read request is forwarded to server.
Step S708, it returns the result, the corresponding data of BootVDI read request is sent to source client.
Step S711, server receives the BootVDI read request that source client is sent, or receives destination client forwarding
BootVDI read request.
Step S712, the corresponding data of server inquiry BootVDI read request whether only existing copy or overload.If BootVDI is read
Corresponding data overload either only existing copy is requested, S713 is thened follow the steps, it is no to then follow the steps S714.
Step S713, addition replacement order.
Step S714, whether inquiry is miss, if so, thening follow the steps S715, otherwise, executes step S716.
Specifically, GCT is safeguarded by server, during client terminal start-up, is pushed to client by server-side.It ran
Server is just broadcasted every certain time interval T in journey updates a GCT.Since renewal time interval may be longer, client
GCT information on end can be expired, at this moment this to receive may need in the client of request without requested cache blocks
Request message is forwarded to server process.Server by forward come request block learn minute mark in the middle part of the GCT of certain client
It records out of date, therefore the neighbouring GCT segment of accessed block is attached in processing result return source client, so that source client
GCT can be updated.In addition, the cache location of server monitoring all clients, the frequent journey that it can change according to client
Degree, adjustment update the frequency of GCT.
Step S715, GCT segment is added, the GCT table of client is updated.
Step S716, server returns the result.
The virtual storage system implementation method of the present embodiment, the method combined by central management and distributed lookup,
Enable each client to read the BootVDI of other clients as needed, reduce the amount of access to server, effect is read in optimization
Rate improves virtual storage system performance；If central server manages each respectively according to the BootVDI forwarding request received
The cooperative caching and GCT table of client, each client can respond the control command of central server, execute distributed pipes
The operation of reason can reduce the expense of server to the greatest extent while guaranteeing the response time.
Fig. 8 shows the structural block diagram of the monitor of virtual machine of one embodiment of the invention.The monitor of virtual machine can be implemented
Above-mentioned virtual storage system implementation method, as shown in figure 8, the monitor of virtual machine 800 specifically includes that
Network unit 802 is used for from server download system virtual disk image BootVDI data, by BootVDI data
It is buffered in a manner of cache blocks in the cooperative caching of monitor of virtual machine.
Read-write cell 803, is connected with network unit, asks for receiving the BootVDI reading from client operating system
It asks, includes the caching block identification of BootVDI data in BootVDI read request.
Processing unit 804 is connected with read-write cell 803 and network unit 802, for will cache in cooperative caching
The cache blocks of BootVDI data are supplied to client operating system.
The monitor of virtual machine of the present embodiment, by being arranged in the VMM of each client for saving system virtualization disk
The cooperative caching of image BootVDI data, and shared cooperative caching, enable each client to read other visitors as needed
The BootVDI at family end, it is possible to reduce to the amount of access of server, efficiency is read in optimization.
In one embodiment, network unit 802 is also used to: the temperature of cache blocks is sent to server；Receive server
The BootVDI data and global buffer of transmission access table GCT, wherein history temperature of the server according to cache blocks, equilibrium assignment
The BootVDI data block that each client need to cache, and allocation result is generated into GCT, the BootVDI number that each client need to be cached
Each client is sent to according to block and GCT.
In one embodiment, processing unit 804 is also used to the history temperature according to each cache blocks, is more than the first threshold by temperature
The cache blocks of value are decomposed at least two cache blocks copies, and make cache blocks copy and cache blocks address having the same.
In one embodiment, processing unit 804 is also used to: being read according to the BootVDI of the client operating system received
Request inquires GCT, determines the address of the cache blocks of BootVDI data；Judge cache BootVDI data cache blocks whether mistake
It carries, and judges whether to hit BootVDI data；If caching the cache blocks overload or miss BootVDI number of BootVDI data
According to transmission BootVDI forwards requests to server.
In one embodiment, network unit 802 is also used to send BootVDI forwarding request to server, wherein server
If receiving BootVDI forwarding request, request is forwarded to return to BootVDI data and control command, control command according to BootVDI
Order is replaced including cache blocks and updates GCT order.
In one embodiment, processing unit 804 is also used to: receiving the BootVDI data block and control life that server returns
It enables, executes control command, by BootVDI data buffer storage in the cache blocks of cooperative caching, replacement temperature is less than second threshold
Cache blocks copy or only existing copy, and update GCT table.
Load balance is introduced cooperative caching by the monitor of virtual machine of the present embodiment, each by DPSB algorithm equilibrium
The temperature of client, so that different clients temperature is uniform as far as possible, each client can read other clients as needed
BootVDI, it is possible to reduce amount of access of the client to server, optimization reading efficiency, raising virtual storage system performance.
Fig. 9 shows the structural block diagram of the virtual storage system server of one embodiment of the invention, which can be implemented
Above-mentioned virtual storage system implementation method, as shown in figure 9, the virtual storage system server 900 specifically includes that
Input-output unit 904, for sending system virtualization disk image BootVDI data, by BootVDI data with slow
The mode of counterfoil is stored in the cooperative caching of the monitor of virtual machine of each client.
Administrative unit 902 is connected with input-output unit 904, for the history temperature according to cache blocks, equilibrium assignment
The BootVDI data block that each client need to cache.
The virtual storage system server of the present embodiment, is arranged cooperative caching in the monitor of virtual machine of client,
By the temperature of the balanced each client of load balancing strategy, so that different clients temperature is uniform as far as possible, each client energy root
According to the BootVDI for needing to read other clients, it is possible to reduce amount of access of the client to server, optimization reading efficiency, raising
Virtual storage system performance.
In one embodiment, administrative unit 902 is also used to the history temperature according to cache blocks, by dynamic equilibrium temperature and
The BootVDI data block that spatial cache block distribution each client of DPSB algorithm equilibrium assignment need to cache, and allocation result is generated
Global buffer accesses table GCT；BootVDI data block and the GCT hair that input-output unit 904 is also used to that each client need to be cached
Give each client.
In this way, by the balanced each client of DPSB algorithm, so that different clients temperature uniformly exists with cache blocks as far as possible
Memory space is occupied in client or the number of cache blocks is uniform as far as possible.
In one embodiment, administrative unit 902 is also used to the history temperature according to cache blocks, is more than first threshold by temperature
Cache blocks be decomposed at least two cache blocks copies, and make cache blocks copy and cache blocks address having the same.
In one embodiment, administrative unit 902 is also used to: if receiving the BootVDI read request of client operating system,
BootVDI data and control command are returned to according to BootVDI read request, control command includes cache blocks replacement order and updates
GCT order.
By being searched in each client distribution formula, if overload or miss, are forwarded to server.
In one embodiment, administrative unit 902 is also used to: if receiving BootVDI forwarding request, being turned according to BootVDI
Hair request returns to BootVDI data and control command, and control command includes that GCT order is ordered and updated in cache blocks replacement.
In this way, each client is managed according to the BootVDI forwarding request received by central server respectively, it can
To reduce the amount of access to server, efficiency is read in optimization, improves virtual storage system performance.
In one embodiment, input-output unit 904 is also used to send BootVDI data and control command to client,
Client executing control command, by BootVDI data-block cache in the cache blocks in cooperative caching, replacement temperature is less than the
The cache blocks copy or only existing copy of two threshold values, and update GCT table.
The virtual storage system server of the present embodiment, is arranged cooperative caching in the monitor of virtual machine of client,
By the temperature of the balanced each client of load balancing strategy, so that different clients temperature is uniform as far as possible, each client energy root
According to the BootVDI for needing to read other clients, it is possible to reduce amount of access of the client to server, optimization reading efficiency, raising
Virtual storage system performance；In addition, passing through the strategy of central server management and distributed lookup, guaranteeing the same of response time
When, the expense of server can be reduced to the greatest extent.
Description of the invention is given for the purpose of illustration and description, and is not exhaustively or will be of the invention
It is limited to disclosed form.Many modifications and variations are obvious for the ordinary skill in the art.It selects and retouches
It states embodiment and is to more preferably illustrate the principle of the present invention and practical application, and those skilled in the art is enable to manage
The solution present invention is to design various embodiments suitable for specific applications with various modifications.
Claims (14)
1. a kind of virtual storage system implementation method characterized by comprising
Monitor of virtual machine in client, will be described from server download system virtual disk image BootVDI data
BootVDI data are buffered in the cooperative caching of the monitor of virtual machine in a manner of cache blocks；The cooperative caching
It is the memory cache in the shared monitor of virtual machine of each client, has been downloaded into the sharing of the client with same requirements
BootVDI data block；
The monitor of virtual machine receives the BootVDI read request from client operating system, in the BootVDI read request
Caching block identification including the BootVDI data；
The monitor of virtual machine inquiry global buffer accesses table GCT, determines the address of the cache blocks of the BootVDI data,
In the case where having cached the cache blocks of the BootVDI data in the cooperative caching, it will delay in the cooperative caching
The cache blocks for the BootVDI data deposited are supplied to the client operating system；
Wherein, the GCT is updated according to the order of server, the frequent degree that the server changes according to client,
Adjustment updates the frequency of GCT；
Wherein, the cache blocks that temperature is more than first threshold are decomposed at least by the server according to the history temperature of cache blocks
Two cache blocks copies, and make the cache blocks copy and cache blocks address having the same.
2. the method according to claim 1, wherein further include:
History temperature of the server according to the cache blocks, the BootVDI data block that each client of equilibrium assignment need to cache,
And allocation result is generated into global buffer and accesses table GCT；
The BootVDI data block and GCT that each client need to cache are sent to each client by the server.
3. according to the method described in claim 2, it is characterized in that, the server according to the history temperatures of the cache blocks,
The BootVDI data block that each client of equilibrium assignment need to cache, comprising:
The server distributes DPSB by dynamic equilibrium temperature and spatial cache block and calculates according to the history temperature of the cache blocks
The BootVDI data block that each client of method equilibrium assignment need to cache.
4. according to the method described in claim 3, it is characterized by further comprising:
The monitor of virtual machine judges whether the cache blocks for caching BootVDI data overload, and judges whether to hit
BootVDI data；
If the cache blocks overload or miss BootVDI data, the monitor of virtual machine that cache BootVDI data are sent
BootVDI forwards requests to server.
5. according to the method described in claim 4, it is characterized by further comprising:
The server receives the BootVDI forwarding request, forwards request to return to BootVDI data according to the BootVDI
And control command, the control command include cache blocks replacement order and update GCT order
And/or
The method also includes:
The monitor of virtual machine receives the BootVDI data and control command that the server returns, and executes control command, will
The BootVDI data buffer storage in the cache blocks of cooperative caching, replacement temperature be less than second threshold cache blocks copy or
Only existing copy, and update the GCT table.
6. a kind of monitor of virtual machine characterized by comprising
Network unit, for from server download system virtual disk image BootVDI data, by the BootVDI data with
The mode of cache blocks caches in the cooperative caching of monitor of virtual machine in the client；The cooperative caching is each client
The memory cache in shared monitor of virtual machine is held, the client with same requirements is made to share the BootVDI number having been downloaded into
According to block；
Read-write cell is connected with the network unit, for receiving the BootVDI read request from client operating system,
It include the caching block identification of the BootVDI data in the BootVDI read request；
Processing unit is connected with the read-write cell and the network unit, for inquiring global buffer access table GCT, really
The address of the cache blocks of the fixed BootVDI data, has cached the caching of the BootVDI data in the cooperative caching
In the case where block, the cache blocks of the BootVDI data cached in the cooperative caching are supplied to the client and are grasped
Make system；
Wherein, the GCT is updated according to the update GCT order of server, the frequency that the server changes according to client
Numerous degree, adjustment update the frequency of GCT；
Wherein, by the server according to the history temperature of cache blocks, by the cache blocks that temperature is more than first threshold be decomposed into
Few two cache blocks copies, and make the cache blocks copy and cache blocks address having the same.
7. monitor of virtual machine according to claim 6, which is characterized in that the network unit is also used to:
The temperature of the cache blocks is sent to the server；
Receive the BootVDI data and global buffer access table GCT that the server is sent, wherein the server is according to institute
The history temperature of cache blocks, the BootVDI data block that each client of equilibrium assignment need to cache are stated, and allocation result is generated into GCT,
BootVDI data block that each client need to cache and the GCT are sent to each client.
8. monitor of virtual machine according to claim 6, which is characterized in that the processing unit is also used to:
Judge whether the cache blocks for caching BootVDI data overload, and judges whether to hit BootVDI data；
If caching the cache blocks overload or miss BootVDI data of BootVDI data, sends BootVDI and forward requests to take
Business device.
9. monitor of virtual machine according to claim 8, which is characterized in that
The network unit is also used to send the BootVDI forwarding request to the server, wherein if the server connects
The BootVDI forwarding request is received, forwards request to return to BootVDI data and control command according to the BootVDI, it is described
Control command includes cache blocks replacement order and update GCT order；
And/or
The processing unit is also used to:
BootVDI data block and control command that the server returns are received, the control command is executed, it will be described
For BootVDI data buffer storage in the cache blocks of cooperative caching, replacement temperature is less than the cache blocks copy or only existing copy of second threshold,
And update the GCT table.
10. a kind of virtual storage system server characterized by comprising
Input-output unit, for sending system virtualization disk image BootVDI data, by the BootVDI data to cache
The mode of block is stored in the cooperative caching of the monitor of virtual machine of each client, and global buffer access table GCT is sent to
Each client；The cooperative caching is the memory cache in the shared monitor of virtual machine of each client, is made with same requirements
Client share the BootVDI data block that has been downloaded into；
Administrative unit is connected with the input-output unit, for the history temperature according to cache blocks, each client of equilibrium assignment
The BootVDI data block that need to be cached is held, and allocation result is generated into the GCT；According to the frequent degree that client changes, adjustment
The frequency for updating GCT sends to each client and updates GCT order；It is more than the first threshold by temperature according to the history temperature of cache blocks
The cache blocks of value are decomposed at least two cache blocks copies, and keep the cache blocks copy and the cache blocks having the same
Location；
Wherein, the monitor of virtual machine receives the BootVDI read request from client operating system, and the BootVDI is read
It include the caching block identification of the BootVDI data in request；The monitor of virtual machine inquiry global buffer accesses table GCT,
The address for determining the cache blocks of the BootVDI data has cached the slow of the BootVDI data in the cooperative caching
In the case where counterfoil, the cache blocks of the BootVDI data cached in the cooperative caching are supplied to the client
Operating system.
11. server according to claim 10, which is characterized in that the administrative unit is also used to according to the cache blocks
History temperature, distribute what each client of DPSB algorithm equilibrium assignment need to cache by dynamic equilibrium temperature and spatial cache block
BootVDI data block；
The input-output unit is also used to BootVDI data block that each client need to cache and the GCT being sent to each visitor
Family end.
12. server according to claim 10, which is characterized in that the administrative unit is also used to:
If receiving the BootVDI read request of client operating system, BootVDI number is returned to according to the BootVDI read request
According to and control command, the control command include cache blocks replacement order and update GCT order；
And/or
The administrative unit is also used to:
If receiving the BootVDI forwarding request, request is forwarded to return to BootVDI data and control according to the BootVDI
Order, the control command include cache blocks replacement order and update GCT order.
13. server according to claim 12, which is characterized in that
The input-output unit is also used to send the BootVDI data to client and control command, the client are held
The row control command, by the BootVDI data-block cache in the cache blocks in cooperative caching, replacement temperature is less than the
The cache blocks copy or only existing copy of two threshold values, and update GCT table.
14. a kind of virtual storage system characterized by comprising
Monitor of virtual machine as described in any in claim 6 to 9；
Virtual storage system server as described in any in claim 10 to 13.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201410497434.7A CN105516223B (en) | 2014-09-25 | 2014-09-25 | Virtual storage system and its implementation, server and monitor of virtual machine |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201410497434.7A CN105516223B (en) | 2014-09-25 | 2014-09-25 | Virtual storage system and its implementation, server and monitor of virtual machine |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN105516223A CN105516223A (en) | 2016-04-20 |
| CN105516223B true CN105516223B (en) | 2019-09-13 |
Family
ID=55723856
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201410497434.7A Active CN105516223B (en) | 2014-09-25 | 2014-09-25 | Virtual storage system and its implementation, server and monitor of virtual machine |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN105516223B (en) |
Families Citing this family (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2018107361A1 (en) * | 2016-12-13 | 2018-06-21 | 华为技术有限公司 | Image processing method and computing device |
| CN106681830B (en) * | 2016-12-21 | 2019-11-29 | 深圳先进技术研究院 | A kind of task buffer space monitoring method and apparatus |
| CN109274550B (en) * | 2018-09-07 | 2020-07-31 | 电信科学技术第五研究所有限公司 | iSCSI self-adaptive IO queue depth matching method |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101419535A (en) * | 2008-11-19 | 2009-04-29 | 北京航空航天大学 | Distributed virtual magnetic disc system of virtual machine |
| CN102387193A (en) * | 2010-09-03 | 2012-03-21 | 马维尔国际贸易有限公司 | Balancing caching load in a peer-to-peer based network file system |
| CN103200245A (en) * | 2013-03-21 | 2013-07-10 | 杭州电子科技大学 | Distributed type network caching method based on Device Mapper |
Family Cites Families (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| KR101459175B1 (en) * | 2010-12-23 | 2014-11-07 | 주식회사 케이티 | System and Terminal Device for Sharing Moving Virtual Images and Method Thereof |
| US20130086579A1 (en) * | 2011-09-30 | 2013-04-04 | Virtual Bridges, Inc. | System, method, and computer readable medium for improving virtual desktop infrastructure performance |
| US9374228B2 (en) * | 2012-10-12 | 2016-06-21 | International Business Machines Corporation | Verifying a geographic location of a virtual disk image executing at a data center server within a data center |
-
2014
- 2014-09-25 CN CN201410497434.7A patent/CN105516223B/en active Active
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101419535A (en) * | 2008-11-19 | 2009-04-29 | 北京航空航天大学 | Distributed virtual magnetic disc system of virtual machine |
| CN102387193A (en) * | 2010-09-03 | 2012-03-21 | 马维尔国际贸易有限公司 | Balancing caching load in a peer-to-peer based network file system |
| CN103200245A (en) * | 2013-03-21 | 2013-07-10 | 杭州电子科技大学 | Distributed type network caching method based on Device Mapper |
Non-Patent Citations (1)
| Title |
|---|
| 《A novel content updating policy based on dynamic storage balancing for clustered streaming media system》;Xing Wei,et al.;《2009 IEEE International Conference on Control and Automation》;20091211;摘要、正文第1-5节 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN105516223A (en) | 2016-04-20 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| US10579272B2 (en) | Workload aware storage platform | |
| CN106790324B (en) | Content distribution method, virtual server management method, cloud platform and system | |
| US8396936B2 (en) | Computer system with cooperative cache | |
| US9330108B2 (en) | Multi-site heat map management | |
| EP2545458B1 (en) | Method and memory cache data center | |
| EP3507694B1 (en) | Message cache management for message queues | |
| US8108620B2 (en) | Cooperative caching technique | |
| US10482062B1 (en) | Independent evictions from datastore accelerator fleet nodes | |
| JP2015072682A (en) | Method and device for load balancing and dynamic scaling of low delay two-layer distributed cache storage system | |
| JP7176209B2 (en) | Information processing equipment | |
| CN108462736B (en) | QoS-oriented cloud storage data copy storage method | |
| US10185666B2 (en) | Item-wise simulation in a block cache where data eviction places data into comparable score in comparable section in the block cache | |
| CN105516223B (en) | Virtual storage system and its implementation, server and monitor of virtual machine | |
| CN110276713A (en) | A kind of high-efficiency caching method and system towards remote sensing image data | |
| CN110377572A (en) | A kind of spatial cache management method, device, equipment and medium | |
| CN110427386A (en) | Data processing method, device and computer storage medium | |
| CN102868542A (en) | Method and system for service quality control in service delivery network | |
| CN110362426B (en) | Selective copy realization method and system for bursty load | |
| CN117539915B (en) | Data processing method and related device | |
| US10841394B2 (en) | Server for caching session information set and method of controlling cache of session information set | |
| JP6793498B2 (en) | Data store device and data management method | |
| Xu et al. | Adaptive and scalable load balancing for metadata server cluster in cloud-scale file systems | |
| JP5661355B2 (en) | Distributed cache system | |
| WO2012136828A1 (en) | Multi-user cache system | |
| WO2017018978A1 (en) | Scheduling jobs in a computing cluster |
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 | ||
| TR01 | Transfer of patent right |
Effective date of registration: 20220125 Address after: 100007 room 205-32, floor 2, building 2, No. 1 and No. 3, qinglonghutong a, Dongcheng District, Beijing Patentee after: Tianyiyun Technology Co.,Ltd. Address before: No.31, Financial Street, Xicheng District, Beijing, 100033 Patentee before: CHINA TELECOM Corp.,Ltd. |
|
| TR01 | Transfer of patent right |