CN108958886A - A kind of virtual machine rapid deployment system and method based on network flow - Google Patents
A kind of virtual machine rapid deployment system and method based on network flow Download PDFInfo
- Publication number
- CN108958886A CN108958886A CN201810708740.9A CN201810708740A CN108958886A CN 108958886 A CN108958886 A CN 108958886A CN 201810708740 A CN201810708740 A CN 201810708740A CN 108958886 A CN108958886 A CN 108958886A
- Authority
- CN
- China
- Prior art keywords
- data
- mirror image
- virtual machine
- cache
- boot
- 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.)
- Granted
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F9/00—Arrangements for program control, e.g. control units
- G06F9/06—Arrangements for program control, e.g. control units using stored programs, i.e. using an internal store of processing equipment to receive or retain programs
- G06F9/44—Arrangements for executing specific programs
- G06F9/455—Emulation; Interpretation; Software simulation, e.g. virtualisation or emulation of application or operating system execution engines
- G06F9/45533—Hypervisors; Virtual machine monitors
- G06F9/45558—Hypervisor-specific management and integration aspects
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F9/00—Arrangements for program control, e.g. control units
- G06F9/06—Arrangements for program control, e.g. control units using stored programs, i.e. using an internal store of processing equipment to receive or retain programs
- G06F9/44—Arrangements for executing specific programs
- G06F9/4401—Bootstrapping
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F9/00—Arrangements for program control, e.g. control units
- G06F9/06—Arrangements for program control, e.g. control units using stored programs, i.e. using an internal store of processing equipment to receive or retain programs
- G06F9/46—Multiprogramming arrangements
- G06F9/50—Allocation of resources, e.g. of the central processing unit [CPU]
- G06F9/5061—Partitioning or combining of resources
- G06F9/5077—Logical partitioning of resources; Management or configuration of virtualized resources
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F9/00—Arrangements for program control, e.g. control units
- G06F9/06—Arrangements for program control, e.g. control units using stored programs, i.e. using an internal store of processing equipment to receive or retain programs
- G06F9/46—Multiprogramming arrangements
- G06F9/54—Interprogram communication
- G06F9/544—Buffers; Shared memory; Pipes
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F9/00—Arrangements for program control, e.g. control units
- G06F9/06—Arrangements for program control, e.g. control units using stored programs, i.e. using an internal store of processing equipment to receive or retain programs
- G06F9/46—Multiprogramming arrangements
- G06F9/54—Interprogram communication
- G06F9/547—Remote procedure calls [RPC]; Web services
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F9/00—Arrangements for program control, e.g. control units
- G06F9/06—Arrangements for program control, e.g. control units using stored programs, i.e. using an internal store of processing equipment to receive or retain programs
- G06F9/44—Arrangements for executing specific programs
- G06F9/455—Emulation; Interpretation; Software simulation, e.g. virtualisation or emulation of application or operating system execution engines
- G06F9/45533—Hypervisors; Virtual machine monitors
- G06F9/45558—Hypervisor-specific management and integration aspects
- G06F2009/45562—Creating, deleting, cloning virtual machine instances
Landscapes
- Engineering & Computer Science (AREA)
- Software Systems (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Computer Security & Cryptography (AREA)
- Stored Programmes (AREA)
- Information Transfer Between Computers (AREA)
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
Abstract
A kind of virtual machine rapid deployment system and method based on network flow, is related to deploying virtual machine field.Time-consuming, can generate a large amount of random I/O request in order to solve the problems, such as the presence of traditional virtual technology by the present invention.In method of the invention, start in deployment, calculate node i starts mirror image transmission, prioritised transmission Boot Cache data, then transmit other parts data in mirror image from other calculate node carry mirror images；Virtual machine instance and corresponding mirror image writable layer are created, mirror image writable layer is linked to mirror image, starts virtual instance；Mirror image is read-only to virtual machine instance, and multiple virtual machine instances share the same mirror image entity, and the data of virtual machine write-in are stored in corresponding mirror image writable layer；Virtual machine start-up course is synchronous with mirror image transmission process to be carried out；After mirror image is transmitted, calculate node i registers mirror cache service to control node, provides buffer service for other calculate nodes.The present invention is used for deploying virtual machine.
Description
Technical field
The present invention relates to deploying virtual machine fields.
Background technique
During the deployment of large-scale virtual network system, the efficiency of mirrored storage and net distribution is to system deployment
Performance has higher influence.Deployment needs to pass through net distribution to each calculate node, so from memory node by mirror image resources
Virtual instance is constructed by calculate node afterwards, and is booted up, service is provided.These mirror image resources sizes are different, from several hundred
All there may be and the networks in deployed environment by MB (simple type linux system) to tens GB (Window Server server)
Bandwidth is often 100Mbps or 1Gbps.Usually, a mirror image guidance is completed, and is typically referred to calculate node and is saved from storage
Point obtains mirror image, profile instance, guides operating system and finally provides service by network.In this process, mirror image is obtained
It is often time-consuming longest.Such as the mirror image of a 4GB size theoretically needs 6 minutes by 100,000,000 networks (100Mbps)
Time.And the guidance of exemplary configuration, operating system tends to complete within 1 minute.In large-scale virtual network portion
It affixes one's name in environment, there is thousands of virtual instance to need to dispose, this problem becomes apparent.
Traditional virtual technology is by disk complete copy to calculate node and then to guide, but virtual machine starts
Required data account for the ratio very little of entire data in magnetic disk, and the demand for realizing that virtual machine quickly guides is not needed entire disk
Data are all transferred to calculate node, a kind of prioritization scheme be it is direct use such as NFS, CIFS Network File System, by local I/O
Request is mapped to network, guarantees that only data in need just pass through network transmission, reduce the data volume of transmission.But virtual power traction
The process of leading can generate a large amount of random I/O request, these requests can bring higher I/O negative to network and long-range memory node
It carries, performance is not outstanding under fairly large virtual instance deployment scenario.
Summary of the invention
Time-consuming, can generate a large amount of random I/O request in order to solve the problems, such as the presence of traditional virtual technology by the present invention.
A kind of virtual machine rapid deployment system based on network flow includes control node and multiple calculate nodes；
The control node include: for the node administration module to control node and each calculate node context management,
For the mirror image management module of Mirror Info management, for monitoring the Load monitoring module of each calculate node load；
The calculate node includes:
The corresponding mirror image writable layer of virtual machine, virtual machine, mirror image, wherein mirror image includes mirror image data caching and file system
Driving unit, VME operating system starting institute must data and mapping table units；
The mirror image data caching and file system driver unit include:
For caching the Local Cache data structure of mirror image data；
The Mount Driver of driving is provided for downloading mirror image data from other calculate nodes for current calculate node；
The VME operating system starting institute data and must map table unit and include:
For cached virtual machine os starting must data BootCache data structure；
For storing the Boot Map of Local Cache data Yu BootCache data mapping relations, i.e. mapping table.
A kind of virtual machine quick deployment method based on network flow, comprising the following steps:
Deployment starts, and calculate node i starts mirror image transmission, prioritised transmission Boot from other calculate node carry mirror images
Cache data, Boot Cache data must data for cached virtual machine os starting institute；Other portions in mirror image are transmitted again
Divided data；
Boot Cache data be according to virtual machine instance starting read sequential delivery, the calculate node i from its
The process of his calculate node carry mirror image data is exactly mirror image transmission process, and mirror image transmission process is synchronous with virtual machine start-up course
It carries out；
Virtual machine instance and corresponding mirror image writable layer are created, mirror image writable layer is linked to mirror image, starts virtual instance；
Mirror image is read-only to virtual machine instance, and multiple virtual machine instances share the same mirror image entity, and the data of virtual machine write-in are stored in
In corresponding mirror image writable layer；
After mirror image is transmitted, calculate node i registers mirror cache service to control node, is other calculate nodes
Buffer service is provided.
Further, the mirror image writable layer uses the format mirror image for supporting Copy on write.
Further, the mirror image transmission process detailed process is as follows:
(1.1) it is directed to calculate node i, creates Boot Cache, Local Cache data structure, initializes drive module
Mount Driver and web socket Socket；
(1.2) new process is created, for from Remote Switched Port Analyzer server downloading Boot Cache data to local, and by data
It is written in Local Cache, is then downloaded in addition to the necessary data of VME operating system starting institute according to mapping relations
Remaining data is into Local Cache；
(1.3) request of virtual machine launching process is monitored, from remote download virtual machine operations system when becoming virtual machine starting
Data except Boot Cache needed for start-up course of uniting, and be written in Local Cache；
(1.4) whether circular test mirror image data downloads completion, until mirror image data downloading is completed, then starts as other meters
Operator node provides image download service.
Further, the virtual machine start-up course, detailed process are as follows:
(2.1) it creates virtual machine instance and corresponding mirror image writable layer, mirror image writable layer is linked in deployment file system
Mirror image；
(2.2) virtual machine sends to mirror image writable layer and requests,
If data with existing in mirror image writable layer, data or write-in data are read, are directly returned；
Otherwise, the data for checking whether cache request in Local Cache, if delayed in Local Cache
The data of request are deposited, the data in Local Cache are read；Otherwise, whether the data for judging request are to deposit in Boot Cache
The data of storage: if it is, process to be transmitted is waited to continue to download, and detect whether corresponding part in Local Cache downloads
It completes, reads data again later；Otherwise, with transmission process communication, the institute from remote download VME operating system start-up course
Data except the Boot Cache needed, and Local Cache is written；
(2.3) this read-write operation is completed, (2.2) is jumped to and carries out next read-write.
Further, described data cached to local from Remote Switched Port Analyzer server downloading Boot Cache, and data are pressed
The process being written in Local Cache according to mapping relations is as follows:
(a) Hook technology is called to record the I/O of VME operating system start-up course, the time comprising request,
Offset of the data of request in disk, the length of the data of request；The corresponding time is requested according to the time after record completion
Sequencing sequence；
(b) each is recorded, according to the sequence that (a) is recorded, the data of corresponding position from virtual machine image file
It once extracts, is stored in a Boot Cache, while recording this mapping relations into mapping table Boot Map.
Beneficial effects of the present invention:
Optimum results of the present invention to mirror image transmission quantity: after the data for needing to transmit 7456MB in conventional transmission scheme,
Operating system can start to start, and the solution of the present invention only needs 153MB, os starting process transmitted data amount drop
Low 97.95%.Optimum results of the present invention to the time: opposite traditional scheme, the present invention reduce 94.63% for time-consuming is disposed,
Relatively remote carry scheme, the present invention reduce 30.51% for time-consuming is disposed.
Detailed description of the invention
Fig. 1 is the structural schematic diagram of the virtual machine rapid deployment system based on network flow；
Fig. 2 is virtual machine Booting sequence figure；
Fig. 3 is the data relationship schematic diagram of Boot Cache and Local Cache；
Fig. 4 is embodiment test flow chart；
Fig. 5 is that booting operating system reads data volume datagram；
Fig. 6 is embodiment test result figure.
Specific embodiment
Specific embodiment 1:
As shown in Figure 1, a kind of virtual machine rapid deployment system based on network flow, includes control node (Controller)
2 and multiple calculate nodes (Compute Node) 1；
The control node includes:
Node for control node and each calculate node context management (information management of the physical configurations such as memory, CPU)
Management module (Node Manager) 23,
For the mirror image management module (Image Manager) 22 of Mirror Info management,
For monitoring the Load monitoring module (Load Monitor) 21 of each calculate node load；
The calculate node includes:
The corresponding mirror image writable layer of virtual machine (VM) 11, virtual machine (R/W Layer) 12, mirror image (Image) 13, wherein mirror
As (Image) includes mirror image data caching and file system driver unit (Image Data), VME operating system starting institute
It must data and mapping table unit (Boot Data)；
Mirror image data caching and file system driver unit include:
Data for caching the Local Cache data structure 133 of mirror image data, in Local Cache data structure
Referred to as Local Cache data；
The Mount of driving is provided for downloading mirror image data from other calculate nodes for current calculate node
Driver134；
VME operating system starting institute must data and map table unit include:
For cached virtual machine os starting must data BootCache data structure 131, BootCache
Data in data structure are known as BootCache data；
The data stored in actually Boot Cache are the read-write operation branch predictions initiated according to virtual machine, with void
There may be differences for data necessary to the quasi- practical start-up course of machine operating system, so in VME operating system start-up course
It may also be desirable to download some random data (in practical starting with the discrepant data of Boot Cache)；
For storing the Boot Map132 of Local Cache data Yu BootCache data mapping relations, i.e. mapping table.
Specific embodiment 2:
Virtual machine quick deployment method based on network flow is based on network flow based on one kind described in specific embodiment one
Virtual machine rapid deployment system realize.
A kind of virtual machine quick deployment method based on network flow, comprising the following steps:
Deployment starts, and calculate node i starts mirror image transmission from other suitable calculate node j carry mirror images, preferential to pass
Defeated Boot Cache data, Boot Cache data must data for cached virtual machine os starting institute；Mirror image is transmitted again
Middle other parts data；
Boot Cache data be according to virtual machine instance starting read sequential delivery, the calculate node i from its
His process of calculate node carry mirror image data (from long-range crawl mirror image data) is exactly mirror image transmission process, mirror image transmission process
It is synchronous with virtual machine start-up course to carry out；
Virtual machine instance and corresponding mirror image writable layer (R/W Layer) are created, mirror image writable layer is linked to mirror image
(Image), start virtual instance；Mirror image (Image) is read-only to virtual machine instance, and multiple virtual machine instances share the same mirror image
The data of entity, virtual machine write-in are stored in corresponding mirror image writable layer；
After mirror image is transmitted, calculate node i registers mirror cache service to control node, is other calculate nodes
Buffer service is provided.
Specific embodiment 3:
Mirror image writable layer described in present embodiment uses the format mirror for supporting Copy on write (COW, copy-on-write)
Picture, including QCOW2 format mirror image.
Other steps are identical with embodiment two.
Specific embodiment 4:
Mirror image transmission process described in present embodiment (process of the calculate node i from other calculate node carry mirror image datas)
As shown in Fig. 2, detailed process is as follows:
(1.1) it is directed to calculate node i, creates Boot Cache, Local Cache data structure, initializes drive module
Mount Driver and web socket Socket；
(1.2) new process is created, for from Remote Switched Port Analyzer server downloading Boot Cache data to local, and by data
It is written in Local Cache, is then downloaded in addition to the necessary data of VME operating system starting institute according to mapping relations
Remaining data is into Local Cache；
It is data cached to be downloaded Boot Cache for creation in transmission process, when downloading, first downloads Boot
Cache is downloaded on one side, on one side according to mapping relations, the data of downloading is copied in Local Cache, in Boot Cache
Data be one piece of continuous data；Being equivalent to Boot Cache is that a part of data in Local Cache individually put forward,
It puts aside, facilitates downloading；In virtual machine start-up course, need to use the data in Boot Cache and Local Cache；
Calculate node can be used as mirror image server after data buffer storage completion and provide mirror image biography for other calculate nodes
Defeated service.
(1.3) request of virtual machine launching process is monitored, from remote download virtual machine operations system when becoming virtual machine starting
Data except Boot Cache needed for start-up course of uniting, and be written in Local Cache；
(1.4) whether circular test mirror image data downloads completion, until mirror image data downloading is completed, then starts as other meters
Operator node provides image download service.
Other steps are identical as specific embodiment two or three.
Specific embodiment 5:
Virtual machine start-up course is as shown in Fig. 2, detailed process is as follows described in present embodiment:
(2.1) it creates virtual machine instance and corresponding mirror image writable layer, mirror image writable layer is linked in deployment file system
Mirror image；
(2.2) virtual machine sends to mirror image writable layer and requests,
If data with existing in mirror image writable layer, data or write-in data are read, are directly returned；
Otherwise, the data for checking whether cache request in Local Cache, if delayed in Local Cache
The data of request are deposited, the data in Local Cache are read；Otherwise, whether the data for judging request are to deposit in Boot Cache
The data of storage: if it is, process to be transmitted is waited to continue to download, and detect whether corresponding part in Local Cache downloads
It completes, reads data again later；Otherwise, with transmission process communication, the institute from remote download VME operating system start-up course
Data except the Boot Cache needed, and Local Cache is written；
(2.3) this read-write operation is completed, (2.2) is jumped to and carries out next read-write.
Other steps are identical as one of specific embodiment two to four.
Specific embodiment 6:
In present embodiment from Remote Switched Port Analyzer server downloading Boot Cache it is data cached to local, and by data according to
The process that mapping relations are written in Local Cache is as follows:
(a) Hook technology is called to record the I/O of VME operating system start-up course, the time comprising request,
Offset of the data of request in disk, the length of the data of request；The corresponding time is requested according to the time after record completion
Sequencing sequence；
(b) as shown in figure 3, being recorded to each, according to (a) record sequence (" 1 ", " 2 " in Fig. 3 ..., " 8 " indicate
Sequentially), the data of corresponding position are once extracted from virtual machine image file, are stored in a Boot Cache,
This mapping relations is recorded into mapping table Boot Map simultaneously.
Other steps are identical as specific embodiment five or six.
Embodiment
Realize that the configuration item of physical host node is as shown in the table:
1 mirror image optimization of system config of table
Client image list:
2 operating system image listing of table
Guest virtual machine is configured to 1 core cpu, 512MB memory.Complete operating system, mirror are mounted in mirror image
Picture format is QCOW2, and virtualization Hypervisor is Libvirt (K (VM)).Operating system is to enter system and actively execute just
Beginningization script, and can lead to the mutual ping of external host to start Success Flag, overall test process is as shown in Figure 4；
The left side is control node deployment process in Fig. 4, and the right is monitoring service procedure.Monitoring service is for recording virtual machine
The time of start completion and relevant information.
After control node starting deployment, recycles to calculate node and send virtual machine enabled instruction, start example, when virtual
After machine start completion, it can be connected to the port that monitoring service is monitored, and send the ID of virtual instance itself to monitoring server,
Deployment process completes backed off after random in all example deployments.Monitoring service processes start simultaneously with deployment process, and starting when records
Time, as initial time.Upon initialization, into circulation is monitored, the connection of virtual instance is waited, it is virtual real to each
The connection of example creates a new process, time when receiving virtual instance ID, and recording connection.
First to I/O requirement analysis, the read-write amount for recording os starting process reading disk is counted, such as Fig. 5 institute
Show；Wherein, the size that the big little finger of toe image file of mirror logic states virtual machine, mirror image actual size refer to image file in place
Practical occupied space in host disk, read request amount refer to the data volume summation of startup stage all read requests, read data
Amount refers to the startup stage practical data volume read from disk.Why read request amount and read data volume it is different, be
Because some data blocks are read in repeatedly in mirror image, it is unwanted that these are duplicate when reading in long-range starting, so will
Data transmission total amount, which is fallen below, reads the numerical value that data volume is shown.
Then image file is set in control node, and starts in calculate node, compared with traditional scheme, performance
As shown in Figure 6；Wherein, local runtime refers to mirror image just in calculate node local, the case where directly starting using Libvirt, routine
Deployment is complete image file to be downloaded to calculate node from control node, and start the time-consuming of virtual machine, and long-range carry is
The image file of control node is directly remotely mounted to calculate node by finger, and starts the time-consuming of virtual machine, and Optimization deployment is exactly
Started with the scheme of this patent.
As can be seen that Conventional deployment scheme be it is very slow, approximately equal to mirror image data is transferred to the time-consuming of calculate node
In addition time of the mirror image in local runtime, long-range carry scheme greatly improves deployment efficiency, but since long-range carry can generate
A large amount of scrappy I/O request, reduces starting speed, and the prioritization scheme that the present invention uses further improves deployment efficiency.
As a comparison, the time-consuming of the scheme of local boot is theoretically the limit of remote deployment time-consuming, and network performance is better, closer to this
A limit.
The present invention is analyzed the I/O process of os starting, and the I/O request data that starting generates is remembered
Record and separation, in deployment this part data in magnetic disk of process prioritised transmission to destination node.On this technology, operating system is opened
The I/O request of dynamic process is analyzed and is recorded, and carries out priority ranking according to sequencing, is started in remote operating system
When, it is transferred to goal systems in the form of data flow, guarantees that os starting and data transmission synchronization carry out, property drops as far as possible
Low latency improves deployment efficiency.
Fig. 5 illustrates the present invention to the optimum results of mirror image transmission quantity, in scheming to 7 operating system mirror image of Windows
For test, the space that disk mirroring distributes in logic is 20480MB (mirror logic size), and image file accounts on host
Space is 7456MB (mirror image actual size), and the total amount of data that os starting process is read from disk mirroring is
289MB (read request amount), some data is identical in this partial data, that is, repeatedly has read identical bits in mirror image
The data set, the data volume of the practical reading in image file are 153MB (reading data volume).
After the data for needing to transmit 7456MB in conventional transmission scheme, operating system can start to start, and this hair
Bright scheme only needs 153MB, and os starting process transmitted data amount reduces by 97.95%.
Fig. 6, which is illustrated, starts the time comparison diagram that a virtual machine needs in different testing schemes, equally with Windows7
For the test of operating system mirror image, traditional scheme is restarted after mirror image is transferred to calculate node from mirror image server, consumption
When 764 seconds, time-consuming 59 seconds by way of long-range carry image file, deployment way through the invention is 41 seconds time-consuming, directly
It is 17 seconds in local boot time-consuming.
It can be seen that opposite traditional scheme, the present invention, which will dispose time-consuming, reduces 94.63%, relatively remote to hang scheme,
The present invention reduces 30.51% for time-consuming is disposed.
Claims (6)
1. a kind of virtual machine rapid deployment system based on network flow includes control node and multiple calculate nodes；
The control node includes: to be used for for the node administration module to control node and each calculate node context management
The mirror image management module of Mirror Info management, for monitoring the Load monitoring module of each calculate node load；
It is characterized in that, the calculate node includes:
The corresponding mirror image writable layer of virtual machine, virtual machine, mirror image, wherein mirror image includes mirror image data caching and file system driver
Unit, VME operating system starting institute must data and mapping table units；
The mirror image data caching and file system driver unit include:
For caching the Local Cache data structure of mirror image data；
The Mount Driver of driving is provided for downloading mirror image data from other calculate nodes for current calculate node；
The VME operating system starting institute data and must map table unit and include:
For cached virtual machine os starting must data Boot Cache data structure；
For storing the Boot Map of Local Cache data Yu Boot Cache data mapping relations, i.e. mapping table.
2. a kind of virtual machine quick deployment method based on network flow, which comprises the following steps:
Deployment starts, and calculate node i starts mirror image transmission, prioritised transmission Boot from other calculate node carry mirror images
Cache data, Boot Cache data must data for cached virtual machine os starting institute；Other portions in mirror image are transmitted again
Divided data；
Boot Cache data are the sequential deliveries read according to virtual machine instance starting, and the calculate node i is counted from other
The process of operator node carry mirror image data is exactly mirror image transmission process, mirror image transmission process and the same stepping of virtual machine start-up course
Row；
Virtual machine instance and corresponding mirror image writable layer are created, mirror image writable layer is linked to mirror image, starts virtual instance；Mirror image
Read-only to virtual machine instance, multiple virtual machine instances share the same mirror image entity, and the data of virtual machine write-in are stored in correspondence
Mirror image writable layer in；
After mirror image is transmitted, calculate node i registers mirror cache service to control node, provides for other calculate nodes
Buffer service.
3. a kind of virtual machine quick deployment method based on network flow according to claim 2, which is characterized in that the mirror image
Writable layer uses the format mirror image for supporting Copy on write.
4. a kind of virtual machine quick deployment method based on network flow according to claim 3, which is characterized in that the mirror image
Transmission process detailed process is as follows:
(1.1) it is directed to calculate node i, creates Boot Cache, Local Cache data structure, initializes drive module Mount
Driver and web socket Socket；
(1.2) create new process, for from Remote Switched Port Analyzer server downloading Boot Cache data to local, and by data according to
Mapping relations are written in Local Cache, then download the residue in addition to the necessary data of VME operating system starting institute
Data are into Local Cache；
(1.3) request of virtual machine launching process is monitored, is opened when becoming virtual machine starting from remote download VME operating system
Data except Boot Cache needed for dynamic process, and be written in Local Cache；
(1.4) whether circular test mirror image data downloads completion, until mirror image data downloading is completed, then starts to calculate section for other
Point provides image download service.
5. a kind of virtual machine quick deployment method based on network flow according to claim 4, which is characterized in that described virtual
Machine start-up course, detailed process are as follows:
(2.1) it creates virtual machine instance and corresponding mirror image writable layer, mirror image writable layer is linked to the mirror in deployment file system
Picture；
(2.2) virtual machine sends to mirror image writable layer and requests,
If data with existing in mirror image writable layer, data or write-in data are read, are directly returned；
Otherwise, the data for checking whether cache request in Local Cache, if buffered in Local Cache ask
The data asked read the data in Local Cache；Otherwise, whether the data for judging request are to store in Boot Cache
Data: if it is, process to be transmitted is waited to continue to download, and detecting whether corresponding part in Local Cache downloads completion,
Read data again later；Otherwise, with transmission process communication, needed for remote download VME operating system start-up course
Data except Boot Cache, and Local Cache is written；
(2.3) this read-write operation is completed, (2.2) is jumped to and carries out next read-write.
6. a kind of virtual machine quick deployment method based on network flow according to claim 4 or 5, which is characterized in that described
It is data cached to local from Remote Switched Port Analyzer server downloading Boot Cache, and data are written to Local according to mapping relations
Process in Cache is as follows:
(a) Hook technology is called to record the I/O of VME operating system start-up course, the time comprising request, request
Offset of the data in disk, the length of the data of request；Corresponding time order and function is requested according to the time after record completion
Sequence sorts；
(b) each is recorded, according to the sequence that (a) is recorded, the data from virtual machine image file corresponding position are primary
It extracts, is stored in a Boot Cache, while recording this mapping relations into mapping table Boot Map.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810708740.9A CN108958886B (en) | 2018-07-02 | 2018-07-02 | Virtual machine rapid deployment system and method based on network flow |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810708740.9A CN108958886B (en) | 2018-07-02 | 2018-07-02 | Virtual machine rapid deployment system and method based on network flow |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN108958886A true CN108958886A (en) | 2018-12-07 |
| CN108958886B CN108958886B (en) | 2022-03-04 |
Family
ID=64484977
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201810708740.9A Active CN108958886B (en) | 2018-07-02 | 2018-07-02 | Virtual machine rapid deployment system and method based on network flow |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN108958886B (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113254081A (en) * | 2021-06-16 | 2021-08-13 | 中国人民解放军国防科技大学 | Mirror image reading and writing system and method for control path in exchange chip |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103870312A (en) * | 2012-12-12 | 2014-06-18 | 华为技术有限公司 | Method and device for establishing storage cache shared by virtual machines |
| CN104050015A (en) * | 2014-06-27 | 2014-09-17 | 国家计算机网络与信息安全管理中心 | Mirror image storage and distribution system for virtual machines |
| JP2015038644A (en) * | 2010-06-30 | 2015-02-26 | 株式会社東芝 | Computer and method of updating virtual machine |
| US9274958B1 (en) * | 2013-02-01 | 2016-03-01 | Parallels IP Holdings GmbH | System and method for cache management in systems that interface to cloud storage |
-
2018
- 2018-07-02 CN CN201810708740.9A patent/CN108958886B/en active Active
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP2015038644A (en) * | 2010-06-30 | 2015-02-26 | 株式会社東芝 | Computer and method of updating virtual machine |
| CN103870312A (en) * | 2012-12-12 | 2014-06-18 | 华为技术有限公司 | Method and device for establishing storage cache shared by virtual machines |
| US9274958B1 (en) * | 2013-02-01 | 2016-03-01 | Parallels IP Holdings GmbH | System and method for cache management in systems that interface to cloud storage |
| CN104050015A (en) * | 2014-06-27 | 2014-09-17 | 国家计算机网络与信息安全管理中心 | Mirror image storage and distribution system for virtual machines |
Non-Patent Citations (2)
| Title |
|---|
| 刘圣卓: "面向虚拟集群的镜像存储与传输优化", 《中国博士学位论文全文数据库 信息科技辑》 * |
| 张钊宁: "云计算大规模弹性资源的性能优化技术研究", 《中国博士学位论文全文数据库 信息科技辑》 * |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113254081A (en) * | 2021-06-16 | 2021-08-13 | 中国人民解放军国防科技大学 | Mirror image reading and writing system and method for control path in exchange chip |
Also Published As
| Publication number | Publication date |
|---|---|
| CN108958886B (en) | 2022-03-04 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| US11556325B2 (en) | Software installation onto a client using existing resources | |
| CN106506587B (en) | Docker mirror image downloading method based on distributed storage | |
| US10193963B2 (en) | Container virtual machines for hadoop | |
| US8495317B2 (en) | System and method for improving performance of data container backups | |
| US10171373B2 (en) | Virtual machine deployment and management engine | |
| US8751785B2 (en) | Memory tagging and preservation during a hot upgrade | |
| US8074038B2 (en) | Converting luns into files or files into luns in real time | |
| US9547605B2 (en) | Method for data backup, device and system | |
| US20120084775A1 (en) | Techniques for Streaming Virtual Machines from a Server to a Host | |
| US9311023B2 (en) | Increased concurrency of an initialization process of multiple data storage units of a volume | |
| US11070419B2 (en) | Methods and systems to troubleshoot and localize storage failures for a multitenant application run in a distributed computing system | |
| US8112620B2 (en) | Method and system for discovery of a root file system | |
| US20140258570A1 (en) | Implementing configuration preserving relocation of sriov adapter | |
| CN108021378A (en) | Upgrade-system, the upgrade method based on virtual machine and device | |
| CN108958886A (en) | A kind of virtual machine rapid deployment system and method based on network flow | |
| CN115408103A (en) | Virtual machine live migration method, system, equipment and storage medium | |
| US9436488B2 (en) | Program redundancy among virtual machines and global management information and local resource information arrangement | |
| Blaas et al. | Stateless provisioning: Modern practice in hpc | |
| US20220405236A1 (en) | Apparatus and method for managing in-memory container storage | |
| CN115509590B (en) | Continuous deployment method and computer equipment | |
| CN107018198A (en) | The method that the DOS systems that a kind of PXE starts share network file | |
| Zheng | This page is NOT part of the dissertation/thesis. It should be disabled before making final, formal submission, but should be included in the version submitted for format check. | |
| CN106155948A (en) | For storing the method for system, aggregation configuration engine, aggregation configuration instrument and file system | |
| CN116860389A (en) | Method and computing device for application recovery of virtual machine | |
| Zheng | The Challenges of Scaling Up High-Throughput Workflow with Container Technology |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |