CN103885829A - Virtual machine cross-data-center dynamic migration optimization method based on statistics - Google Patents
Virtual machine cross-data-center dynamic migration optimization method based on statistics Download PDFInfo
- Publication number
- CN103885829A CN103885829A CN201410153740.9A CN201410153740A CN103885829A CN 103885829 A CN103885829 A CN 103885829A CN 201410153740 A CN201410153740 A CN 201410153740A CN 103885829 A CN103885829 A CN 103885829A
- Authority
- CN
- China
- Prior art keywords
- memory pages
- data
- virtual machine
- data center
- page
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Images
Abstract
A virtual machine cross-data-center dynamic migration optimization method based on statistics includes the implementing steps that based on the statistics of virtual machine internal storage pages of a data center, a template page is selected, and an index is built; in a source data center, if an internal storage page to be transmitted is the same as the template page, only the index is transmitted, and internal storage page data are recovered again in a target end data center according to the index; if the internal storage page to be transmitted is different from the template page, the internal storage page data are transmitted to the target end data center. Storage resource expenditure in the migration process is effectively reduced, the data transmission process is accelerated, and the virtual machine cross-data-center dynamic migration optimization method has remarkable performance advantages in virtual machine cross-data-center dynamic migration.
Description
Technical field
The present invention relates to cloud computing and virtual field, be specifically related to a kind of virtual machine (vm) migration optimization method based on statistics.
Background technology
Virtual machine dynamic migration (Live Migration) refers to that the operating system based on virtual machine operation can be moved to another physical machine from a physical machine incessantly moves, and this technology is widely used under the scenes such as the on-line maintenance of load balancing, managing power consumption, System Fault Tolerance and software and hardware.Virtual machine dynamic migration is as one of virtualized core key technology, and for setting up, capable of dynamic is flexible, the large-scale green data center of resource intensive is significant.
Virtual machine, across the dynamic migration of data center, need to copy the internal storage state of virtual machine to destination data center from source data center completely, keeps the operation of virtual machine simultaneously.Than data center inside, between data center, communication delay message transmission rate large, that can be used for migration is lower, and internal storage data changes comparatively fast, and this has brought larger difficulty to realizing virtual machine across data center's dynamic migration.And, in the scene across data center's migration, often needing the Virtual Cluster (Virtual Cluster) of several virtual machine compositions to do as a whole migration, this has more increased the difficulty of dynamic migration.
At present, internal memory pre-copy dynamic migration method is the method generally adopting, but it has moved the internal storage data of a large amount of repetitions to destination data center from source data center, and this needs extra transit time and downtime.Although the network service speed between data center will be more and more higher, the scale of cluster is also increasing rapidly, so solve virtual machine across data center's dynamic migration problem, can not rely on merely the raising of network rate.Solving virtual machine is to reduce the data volume of transmitting in copy procedure across the key of data center's migration internal storage state copy.Owing to having moved identical or similar software in different virtual machine, so there are a large amount of identical memory pages between the different virtual machine of data center, this makes to adopt data duplicate removal (Data Deduplication) method to reduce the data volume of transmitting in virtual machine dynamic migration process.
Data duplicate removal is a kind of by the extensive redundant data of eliminating, and reduces the important technology of data carrying cost.Repeating data is present in the links that virtual machine information is processed and stored in a large number, as file system, file synchronization, Email attachment, html document and Web object, and in operating system and application software, migration repeating data can be wasted a large amount of network bandwidths and storage space.Data duplicate removal technology can not only reduce the memory space of data, reduces the expense of storage resources, reduces the cost that administers and maintains of physical store resource, also can save the network bandwidth, speeding up data transmission process.
The existing mode of utilizing data duplicate removal technology to improve virtual machine dynamic migration performance is mainly divided into two kinds.The first is to utilize memory pages internal features to carry out duplicate removal.The acquiescence moving method of for example QEMU-KVM, referring to A.Kivity, Y.Kamay, D.Laor; U.Lublin, and A.Liguori, " Kvm:The linux virtual machine monitor; " In Proceedings of Linux Symposium, 2007, pp.225-230.; For example, when all bytes of memory pages are all when identical (zero page face), only need the data of a byte of transmission and avoid whole memory pages content to transmit.This method deduplicated efficiency is high, recover simple, but duplicate removal probability is very low, because meet the page limited amount of this duplicate removal condition in memory pages set.Therefore this method is very limited to the lifting of virtual machine dynamic migration performance.
Second method is to utilize the feature that different memory pages contents are identical to carry out duplicate removal, for example Shrinker method, P.Riteau, C.Morin, and T.Priol, " Shrinker:Improving live migration of virtual clusters over wans with distributed data deduplication and content-based addressing, " In Proceedings of EUROPAR, September2011..This method is got off the page cache having transmitted index building, when the subsequent transmission page carries out duplicate removal when the page of buffer memory repeats, replaces and transmits complete memory pages content with transmission index value.This method duplicate removal probability is higher, still, due to the defect of buffer memory itself, in the time moving a virtual machine or move many isomery virtual machines, there will be hit rate low, the problem of dynamic migration poor performance.
Summary of the invention
Technology of the present invention is dealt with problems: overcome the deficiencies in the prior art, a kind of optimization method that is applicable to carry out across data center virtual machine dynamic migration has been proposed in the present invention, the method is based on statistics, extract memory pages and analyze from a large amount of computing machines that moving operating system and load, adopt data duplicate removals (Data Deduplication) method to reduce the data volume of transmitting in virtual machine dynamic migration process, effectively reduce the expense of the storage resources in transition process, accelerate data transmission procedure, virtual machine across data center's dynamic migration on there is significant performance advantage.
The technology of the present invention solution: the virtual machine based on statistics, across data center's dynamic migration optimization method, said method comprising the steps of:
(1) number of times that statistics virtual machine occurs across data center's same memory page, chooses template page, builds memory pages index;
(2) virtual machine, across in data center's dynamic migration process, at source data center, if memory pages to be transmitted is identical with template page, only transmits this memory pages index; If memory pages to be transmitted is different from template page, transmit this memory pages data;
(3) virtual machine is across in data center's dynamic migration process, in destination data center, if destination virtual machine receives is memory pages index, navigates to memory pages data group according to memory pages index and recovers memory pages data; If what destination virtual machine received is memory pages data, directly give corresponding virtual machine space by this memory pages data transmission.
The method of choosing template page in described step (1) is as follows: the frequency n >=N occurring in across data center at virtual machine as a certain memory pages tp, wherein, N is the constant of a setting, claims that memory pages tp is template page, and the set of all template pages is denoted as TP.
The process that described step (1) builds memory pages index is as follows:
(1) use hash function to ask for the cryptographic hash of memory pages;
(2) according to cryptographic hash and memory pages content one to one mode store data in a table with the sequence of cryptographic hash size.
Described step (2) is implemented as follows:
(21) use hash function to ask for the cryptographic hash of memory pages to be transmitted;
(22) cryptographic hash of searching memory pages to be transmitted in the array of cryptographic hash of storing all template pages;
(23) if search and hit, the cryptographic hash of actual transmissions memory pages to be transmitted, does not hit if searched, actual transmissions memory pages content to be transmitted.
Described step (3) is implemented as follows:
(31) destination main frame receives after migration data, calculates the address of this memory pages according to the head of data flow architecture, and receives the type of data according to the zone bit judgement in this head;
(32) if judging what receive is memory pages cryptographic hash, in the table of storing template content of pages and cryptographic hash, recovers memory pages content and transfer data to the assigned address that previous step calculates; If judging what receive is memory pages content, the assigned address directly data transmission receiving being calculated to previous step.
The present invention's beneficial effect is compared with prior art:
(1) the present invention is directed to virtual machine across in data center's dynamic migration process, wan communication postpones large, low for the message transmission rate of moving, and the contradiction between internal storage data renewal speed is fast, extract memory pages and analyze from a large amount of computing machines that moving operating system and load, adopt data duplicate removals (Data Deduplication) method to reduce the data volume of transmission, based on the statistics of data center's virutal machine memory page, choose template page index building, at source data center, if memory pages to be transmitted is identical with template page, only transmit index, recover memory pages data according to index again in destination data center, if memory pages to be transmitted is different from template page, by this memory pages data transmission to destination data center, this method has effectively reduced the expense of the storage resources in transition process, accelerated data transmission procedure, virtual machine across data center's dynamic migration on there is significant performance advantage.
(2) the present invention all can effectively reduce transmitted data amount in the first run iteration of internal storage state and successive iterations copy procedure, thereby significantly improves the migration performance of virtual machine.
Brief description of the drawings
Fig. 1 is that virtual machine in the inventive method is across data center's dynamic migration process;
Fig. 2 is the data flow architecture in migration;
Fig. 3 index structuring method process flow diagram of the present invention.
Embodiment
Between the different virtual machine of data center, there are a large amount of identical memory pages, repeat memory pages.To produce because having moved identical or similar software in virtual machine owing to repeating memory pages, so the frequency that repetition memory pages occurs in same data center is also higher, the number of times occurring by the statistics center same memory page, can find the memory pages that repetition rate is high.At this, definition template page is as follows:
Definition: as frequency n >=N that a certain memory pages tp occurs in data center, wherein, N is the constant of a setting, claims that memory pages tp is template page, and the set of all template pages is denoted as TP.
Technical scheme of the present invention is as follows:
Based on the statistics of data center's virutal machine memory page, choose template page index building.At source data center, if memory pages to be transmitted is identical with template page, only transmit index, recover memory pages data according to index again in destination data center; If memory pages to be transmitted is different from template page, by this memory pages data transmission to destination data center.
Virtual machine across data center's dynamic migration process as shown in Figure 1.Said method comprising the steps of:
(1) source node and destination node are held consultation, and connect and start to move virtual machine.
(2) the index h of source data center construction memory pages p to be transmitted, and search h in array of indexes H, to determine that whether p is identical with a certain template page tp.
(3) if p is identical with a certain template page tp, index h is sent to destination data center; If p is different from all template pages, memory pages data p is sent to destination data center.
(4) if destination virtual machine receives is index h, in concordance list HT, find a F (th, tp), wherein th=h, navigates to the some concrete memory pages in memory pages data array, is transferred to corresponding virtual machine space; If what destination virtual machine received is memory pages data, be directly transferred to corresponding virtual machine space.
(5) migration finishes, and stops the operation of virtual machine on source node, starts to move virtual machine on destination node, and disconnects.
Implementing platform needs a complete KVM running environment, and KVM is the abbreviation based on kernel virtual machine, is the solution of the Full-virtualization of hardware supported.Except needing at kernel spacing KVM module, also need QEMU to simulate required CPU and device model and for starting client process at user's space.Qemu-kvm is the QEMU branch in order to have made special amendment and optimization for KVM, and it supports four kinds of migration models, is respectively tcp pattern, exec pattern, unix pattern and fd pattern.What implementation algorithm of the present invention used is tcp pattern, realizes virtual machine dynamic migration by the mode of tcp connection transmission internal storage data.
In Fig. 2, addr field before page data is to calculate the required data of memory address, wherein first and second and four at end are used to markup page type, wherein RAM_SAVE_FLAG_FULL is used for representing the template duplicate removal page (RAM_SAVE_FLAG_DUPLICA), RAM_SAVE_FLAG_PAGE is used for representing ordinary pages, and RAM_SAVE_FLAG_COMPRESS is used for representing that the page that all bytes are consistent is as zero page face.As shown in Figure 1, 2, the specific embodiment of the invention mainly comprises following step:
1. index building
(11) source node and destination node are held consultation, and connect and start to move virtual machine.
(12) the index h of source data center construction memory pages p to be transmitted, and search h in array of indexes H, to determine that whether p is identical with a certain template page tp.
(13) if p is identical with a certain template page tp, index h is added into the page data part of data stream, and page marks is set to RAM_SAVE_FLAG_FULL, send to destination data center; If p is different from all template pages, the memory pages data p that is labeled as RAM_SAVE_FLAG_PAGE is sent to destination data center.
(14) if destination virtual machine receives is index h, in concordance list HT, find a F (th, tp), wherein th=h, navigates to the some concrete memory pages in memory pages data array, is transferred to corresponding virtual machine space; If what destination virtual machine received is memory pages data, be directly transferred to corresponding virtual machine space.
After above step executes, migration finishes, and stops the operation of virtual machine on source node, starts to move virtual machine on destination node, and disconnects.
The memory pages data of the virtual machine selecting from data center by statistical study, obtain template page, then these memory pages are seeked and are drawn successively, and build a Bloom filter structure taking all indexes as key, memory-resident.
Present embodiment is carried out index building as an example of the hash function of SHA-1 example.Hash function (Hash function, hash function) is one of effective ways to data index building.The selection of hash function need to meet anti-collision (Collision Resistance) requirement, and it is mainly the different mistakes that produce of memory pages that the memory pages in order to prevent from recovering according to cryptographic hash in destination data center sends from source data center that the anti-collision here requires.This implementation process is selected the SHA-1 hash function that cryptographic hash length is 160bits, may need to represent with 192 integers in real system.This hash function bumps and the probability occurring that leads to errors, and therefore the probability occurring far below TCP error of transmission or Installed System Memory mistake, collides by cryptographic hash the mistake causing negligible.Meanwhile, SHA-1 function has now the hardware that very ripe speed is very high and realizes, and can not cause transmission delay or reduce transfer rate because of the computing cost of SHA-1.
Index building complete procedure is as follows:
First, choose at random some virtual machines from data center, its internal storage state is preserved to snapshot, obtain m memory pages, the set that this m memory pages formed is denoted as M;
Then, according to Figure 3 shows that the statistical mask page index building, the corresponding tlv triple E=(h of each template page, c, p), the data that wherein p is this memory pages, h=SHA-1 (p), c is the number of times that this memory pages occurs in set M; Shown in algorithm flow chart, be input as memory pages set M and template page decision threshold N, be output as array H and Hash table HT, the size of H and HT is ct, is made up of two parts:
(1) process successively memory pages set M, build binary sort tree T, each node of T is tlv triple E.Add up the nodes that c is greater than N simultaneously, obtain ct.
(2) root binary tree traversal T, the node that c is greater than to N is inserted H and HT.
Finally, the array H that comprises template page cryptographic hash is sent to source data center, the Hash table HT that comprises template page data is kept to destination data center.Array H arranges from small to large ord, and the element of array HT is arranged from small to large according to cryptographic hash.
When source data center starts operation, the cryptographic hash of whole template pages is deposited in an array memory-resident, then build a structure array, in structure, comprise the cryptographic hash deposited in array and corresponding array index, according to cryptographic hash size to this structure array sort; When destination data center starts, also deposit the content of whole template pages in an array memory-resident, memory pages content array and memory pages cryptographic hash array are by array index one to one.
2. source processing procedure
(21) source starts migration by the order of qemu-kvm.KVM acquiescence migration pattern has the memory pages of two types, be all bytes of the memory pages memory pages that all byte is not quite similar in identical memory pages (overwhelming majority is zero page face) and the page, when this two kinds of memory pages transmission, have corresponding flag byte.In the time of migration, source virtual machine is asked cryptographic hash to memory pages data to be transmitted (flag byte is designated common memory pages).Then the cryptographic hash of this memory pages is carried out to binary chop in the array after sequence.If find this cryptographic hash, again stamp a kind of duplicate removal page marker (RAM_SAVE_FLAG_DUPLICA) to memory pages, be index value by 4k memory pages content replacement, use socket to be transferred to destination, if do not found, do not change memory pages data, and it is transferred to destination from source virtual machine.
(22), while moving virtual machine, its data are not stopped transmission with the form of data stream.The structure of data stream as shown in Figure 2.Data stream is made up of data cell, and each data cell is made up of 64 potential head portions and memory pages data.Data cell contains memory pages data, and the head of memory pages data has the zone bit of 12.The 9th and the 11st is for showing whether migration memory pages takes the QEMU/KVM duplicate removal method of acquiescence to compress.If the 11st is set to 1, illustrate that memory pages is compressed; Otherwise, be normal memory pages, the zone bit of the 9th will be set to 1 simultaneously.Zone bit last position, in QEMU-KVM-1.2.0, put aside as useless for the 12nd, thus with it identify duplicate removal memory pages, RAM_SAVE_FLAG_DUPLICA is called after RAM_SAVE_FLAG_FULL in source code.For such data cell, the storage data length of data cell is 20Byte, i.e. the cryptographic hash of this memory pages.In virtual machine, the side-play amount of memory pages can be calculated by 64 potential head portions.
3. destination processing procedure
(31) before migration starts, the virtual machine monitor of destination creates a virtual machine space and is used for placing virtual machine and opens a node and monitor migration request, sets up TCP be connected with source data center simultaneously.When data arrive destination, destination moves passive receive.
(32) destination receives the flag byte that first judges memory pages after data, all identical memory pages of the common memory page or byte if, the corresponding virtual machine space that directly sends to migration order to create.Repeat if memory pages (being labeled as RAM_SAVE_FLAG_DUPLICA), the RAM_SAVE_FLAG_DUPLICA zone bit of the data cell receiving is set at 1 o'clock, take out two bytes that represent thereafter array index, navigate to rapidly the some concrete memory pages in memory pages data array, then the flag byte of this memory pages is set to commonly, is transferred to corresponding virtual machine space.
(33), in the case of expecting that the transmission time is less than the maximum downtime of setting, source virtual machine is hung up, and the state of the free memory page and CPU, register is transferred to destination in the lump, destination virtual machine activation, and migration completes.
Non-elaborated part of the present invention belongs to techniques well known.
The above; be only part embodiment of the present invention, but protection scope of the present invention is not limited to this, in the technical scope that any those skilled in the art disclose in the present invention; the variation that can expect easily or replacement, within all should being encompassed in protection scope of the present invention.
Claims (4)
1. the virtual machine based on statistics, across data center's dynamic migration optimization method, is characterized in that: said method comprising the steps of:
(1) number of times that statistics virtual machine occurs across data center's same memory page, chooses template page, builds memory pages index; The method of choosing described template page is as follows: as frequency n >=N that a certain memory pages tp occurs in across data center at virtual machine, wherein, N is the constant of a setting, claims that memory pages tp is template page, and the set of all template pages is denoted as TP;
(2) virtual machine, across in data center's dynamic migration process, at source data center, if memory pages to be transmitted is identical with template page, only transmits this memory pages index; If memory pages to be transmitted is different from template page, transmit this memory pages data;
(3) virtual machine is across in data center's dynamic migration process, in destination data center, if destination virtual machine receives is memory pages index, navigates to memory pages data group according to memory pages index and recovers memory pages data; If what destination virtual machine received is memory pages data, directly give corresponding virtual machine space by this memory pages data transmission.
2. the virtual machine based on statistics according to claim 1, across data center's dynamic migration optimization method, is characterized in that: the process that described step (1) builds memory pages index is as follows:
(1) use hash function to ask for the cryptographic hash of memory pages;
(2) according to cryptographic hash and memory pages content one to one mode store data in a table with the sequence of cryptographic hash size.
3. the virtual machine based on statistics according to claim 1, across data center's dynamic migration optimization method, is characterized in that: described step (2) is implemented as follows:
(21) use hash function to ask for the cryptographic hash of memory pages to be transmitted;
(22) cryptographic hash of searching memory pages to be transmitted in the array of cryptographic hash of storing all template pages;
(23) if search and hit, the cryptographic hash of actual transmissions memory pages to be transmitted, does not hit if searched, actual transmissions memory pages content to be transmitted.
4. the virtual machine based on statistics according to claim 1, across data center's dynamic migration optimization method, is characterized in that: described step (3) is implemented as follows:
(31) destination main frame receives after migration data, calculates the address of this memory pages according to the head of data flow architecture, and receives the type of data according to the zone bit judgement in this head;
(32) if judging what receive is memory pages cryptographic hash, in the table of storing template content of pages and cryptographic hash, recovers memory pages content and transfer data to the assigned address that previous step calculates; If judging what receive is memory pages content, the assigned address directly data transmission receiving being calculated to previous step.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201410153740.9A CN103885829A (en) | 2014-04-16 | 2014-04-16 | Virtual machine cross-data-center dynamic migration optimization method based on statistics |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201410153740.9A CN103885829A (en) | 2014-04-16 | 2014-04-16 | Virtual machine cross-data-center dynamic migration optimization method based on statistics |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN103885829A true CN103885829A (en) | 2014-06-25 |
Family
ID=50954738
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201410153740.9A Pending CN103885829A (en) | 2014-04-16 | 2014-04-16 | Virtual machine cross-data-center dynamic migration optimization method based on statistics |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN103885829A (en) |
Cited By (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104156254A (en) * | 2014-07-29 | 2014-11-19 | 北京高森明晨信息科技有限公司 | Method and device for migrating virtual machines |
| CN105468457A (en) * | 2015-11-24 | 2016-04-06 | 无锡江南计算技术研究所 | Parallel system local migration fault-tolerant method based on difference recognition |
| US9898319B2 (en) | 2015-02-12 | 2018-02-20 | National Central University | Method for live migrating virtual machine |
| CN108469982A (en) * | 2018-03-12 | 2018-08-31 | 华中科技大学 | A kind of online moving method of container |
| CN109684829A (en) * | 2018-12-04 | 2019-04-26 | 中国科学院数据与通信保护研究教育中心 | Service call monitoring method and system in a kind of virtualized environment |
| US10691692B2 (en) | 2016-04-29 | 2020-06-23 | Fujitsu Limited | Computer-implemented method of executing a query in a network of data centres |
| CN112416951A (en) * | 2019-08-21 | 2021-02-26 | 北京京东乾石科技有限公司 | Data transmission method and device, computer storage medium |
| CN112948052A (en) * | 2021-02-23 | 2021-06-11 | 中国联合网络通信集团有限公司 | Cross-data-center virtual machine migration method, data center and computer medium |
| US11074092B2 (en) | 2015-12-18 | 2021-07-27 | Intel Corporation | Virtual machine batch live migration |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN1897570A (en) * | 2006-06-15 | 2007-01-17 | 清华大学 | Large data-set task node transferrying method based on content and relation separation |
| CN103034591A (en) * | 2012-11-21 | 2013-04-10 | 北京航空航天大学 | Memory sharing method and device for virtual machine |
| CN103581331A (en) * | 2013-11-13 | 2014-02-12 | 中国科学院计算技术研究所 | Virtual machine on-line transfer method and system |
-
2014
- 2014-04-16 CN CN201410153740.9A patent/CN103885829A/en active Pending
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN1897570A (en) * | 2006-06-15 | 2007-01-17 | 清华大学 | Large data-set task node transferrying method based on content and relation separation |
| CN103034591A (en) * | 2012-11-21 | 2013-04-10 | 北京航空航天大学 | Memory sharing method and device for virtual machine |
| CN103581331A (en) * | 2013-11-13 | 2014-02-12 | 中国科学院计算技术研究所 | Virtual machine on-line transfer method and system |
Non-Patent Citations (1)
| Title |
|---|
| MINGYU LI ET AL.: "Template-based Memory Deduplication Method for Inter-Data Center Live Migration of Virtual Machines", <<CLOUD ENGINEERING(IC2E),2014 IEEE INTERNATIONAL CONFERENCE ON>> * |
Cited By (14)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104156254A (en) * | 2014-07-29 | 2014-11-19 | 北京高森明晨信息科技有限公司 | Method and device for migrating virtual machines |
| US9898319B2 (en) | 2015-02-12 | 2018-02-20 | National Central University | Method for live migrating virtual machine |
| CN105468457A (en) * | 2015-11-24 | 2016-04-06 | 无锡江南计算技术研究所 | Parallel system local migration fault-tolerant method based on difference recognition |
| CN105468457B (en) * | 2015-11-24 | 2019-04-09 | 无锡江南计算技术研究所 | A kind of parallel system local migration fault-tolerance approach based on difference identification |
| US11074092B2 (en) | 2015-12-18 | 2021-07-27 | Intel Corporation | Virtual machine batch live migration |
| US10691692B2 (en) | 2016-04-29 | 2020-06-23 | Fujitsu Limited | Computer-implemented method of executing a query in a network of data centres |
| CN108469982A (en) * | 2018-03-12 | 2018-08-31 | 华中科技大学 | A kind of online moving method of container |
| CN108469982B (en) * | 2018-03-12 | 2021-03-26 | 华中科技大学 | Online migration method for container |
| CN109684829B (en) * | 2018-12-04 | 2020-12-04 | 中国科学院数据与通信保护研究教育中心 | Service call monitoring method and system in virtualization environment |
| CN109684829A (en) * | 2018-12-04 | 2019-04-26 | 中国科学院数据与通信保护研究教育中心 | Service call monitoring method and system in a kind of virtualized environment |
| CN112416951A (en) * | 2019-08-21 | 2021-02-26 | 北京京东乾石科技有限公司 | Data transmission method and device, computer storage medium |
| CN112416951B (en) * | 2019-08-21 | 2024-03-01 | 北京京东乾石科技有限公司 | Data transfer method and device and computer storage medium |
| CN112948052A (en) * | 2021-02-23 | 2021-06-11 | 中国联合网络通信集团有限公司 | Cross-data-center virtual machine migration method, data center and computer medium |
| CN112948052B (en) * | 2021-02-23 | 2023-10-13 | 中国联合网络通信集团有限公司 | Virtual machine migration method across data centers, data center and computer medium |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN103885829A (en) | Virtual machine cross-data-center dynamic migration optimization method based on statistics | |
| US10761933B2 (en) | Prefill of raid stripes in a storage system by reading of existing data | |
| CN103473277B (en) | The Snapshot Method and device of file system | |
| CN102307206B (en) | Caching system and caching method for rapidly accessing virtual machine images based on cloud storage | |
| CN103765373B (en) | Date storage method, data storage device and memory device | |
| CN102521063B (en) | Shared storage method suitable for migration and fault tolerance of virtual machine | |
| CN111124277B (en) | Deep learning data set caching method, system, terminal and storage medium | |
| CN110262922B (en) | Erasure code updating method and system based on duplicate data log | |
| CN102457283B (en) | A kind of data compression, decompression method and equipment | |
| CN104731569B (en) | A kind of data processing method and relevant device | |
| CN102663058A (en) | URL duplication removing method in distributed network crawler system | |
| EP3316150A1 (en) | Method and apparatus for file compaction in key-value storage system | |
| CN114201421B (en) | Data stream processing method, storage control node and readable storage medium | |
| Zhang et al. | Smarth: Enabling multi-pipeline data transfer in hdfs | |
| CN110347651A (en) | Method of data synchronization, device, equipment and storage medium based on cloud storage | |
| CN103246484B (en) | A kind of date storage method, Apparatus and system | |
| CN103605630A (en) | Virtual server system and data reading-writing method thereof | |
| CN111026510A (en) | Cross-storage thermal migration method for virtual machine in cloud environment | |
| CN111966631A (en) | Mirror image file generation method, system, equipment and medium capable of being rapidly distributed | |
| Senger et al. | BSP cost and scalability analysis for MapReduce operations | |
| CN106254270A (en) | A kind of queue management method and device | |
| CN106156142A (en) | The processing method of a kind of text cluster, server and system | |
| CN113900810A (en) | Distributed graph processing method, system and storage medium | |
| CN111290714B (en) | Data reading method and device | |
| CN112328697A (en) | Data synchronization method based on big data |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| WD01 | Invention patent application deemed withdrawn after publication |
Application publication date: 20140625 |
|
| WD01 | Invention patent application deemed withdrawn after publication |