CN103077055A - Method for high-efficiently supporting online starting and running of plenty of virtual machines through parallel network file system (pNFS) system - Google Patents
Method for high-efficiently supporting online starting and running of plenty of virtual machines through parallel network file system (pNFS) system Download PDFInfo
- Publication number
- CN103077055A CN103077055A CN2012105926290A CN201210592629A CN103077055A CN 103077055 A CN103077055 A CN 103077055A CN 2012105926290 A CN2012105926290 A CN 2012105926290A CN 201210592629 A CN201210592629 A CN 201210592629A CN 103077055 A CN103077055 A CN 103077055A
- Authority
- CN
- China
- Prior art keywords
- file
- data server
- pnfs
- client
- virtual machine
- 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
Images
Landscapes
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
- Stored Programmes (AREA)
Abstract
The invention relates to the cloud computing technology, and particularly discloses a method for high efficiently supporting online starting and running of plenty of virtual machines through a parallel network file system (pNFS) system. The method comprises the following steps that a client acquires the information of the layout type of the pNFS system from a metadata server and judges whether the layout type of the pNFS system is compatible with the client or not, and enters a next step if the pNFS system is compatible with the client; the client requests the layout information of a needed file to the metadata server, and the metadata server immediately calculates the layout information of the file and transmits the layout information to the client; and the client stores the file data to a corresponding data server in parallel according to the layout information of the file. Due to the adoption of the technical scheme, the metadata information of a minimum discernible signal (MDS) end can be simplified to alleviate the pressure on the MDS when a great number of files are transmitted and read in parallel, the DS (data server) data distribution can be dynamically balanced, the file data is more uniformly distributed to each DS, and the method is particularly applicable to the online starting and running of a great number of virtual machines under a general scene.
Description
Technical field
The present invention relates to the cloud computing technology field, particularly a kind of pNFS system high efficiency supports the magnanimity virtual machine to start online the method for operation.
Background technology
In recent years, along with the develop rapidly of computing machine and Internet technology, increasingly huge, a large amount of physical resource of various IT resources comprises that the data resource of computational resource, Service Source, software resource and magnanimity is by network polymerization.How to effectively utilize these physical resources, improve the efficient utilization of resource rate and but perplexing the Resource Owner always.In order to satisfy the growing demand for the large-scale calculations ability of all kinds of complicated applications, how on these physical resources, come the virtual computation environmental that meets the demands for user's rapid build according to user's demand, improve the comprehensive utilization of resources rate, just naturally become the major issue that current urgent need solves.
Risen rapidly a kind of new commercial computation model---" cloud computing (Cloud Computing) " beginning of this century.It consists of a huge resource pool with the computing machine of magnanimity by network interconnection, manages these resources by specific software systems, uses for the user provides as required, by the Highly Scalable service that makes paying.The cloud computing frame of reference composition that whole world IT giant Microsoft proposes is comprised of resource layer, virtualization layer, administration and supervision authorities and service layer's four parts.Resource layer is comprised of the various isomery physical resources of bottom, and these physical resources can distribute and withdrawal neatly by after virtualization layer virtual, so that administration and supervision authorities are carried out unified management easily, and provides possibility for monitoring resource and load balance.Service layer makes things convenient for the user effectively to satisfy the application demand of self with the physical resource of bottom for the user provides some cloud computings relevant service.Wherein the correlation technique of virtualization layer be realize cloud computing most important also be to study a widest content.The solution of cloud computing be unable to do without virtual, Intel Virtualization Technology is isolating hardware and software effectively, it provides solid foundation to cloud computing, how to distribute virtual resource to provide better service for the user so that cloud computing only need to be concerned about, and needn't remove directly to manage the physical resource of bottom.
The current Intel Virtualization Technology route that exists two kinds of main flows: fully virtualized (full-virtualization) and half virtual (para-virtualization).Fully virtualized take VMware as representative, its advantage is that operating system can just directly be moved in physical machine without any modification; But simultaneously its binary translation technology to the part privileged instruction of adopting also can be brought certain system performance expense.Half virtual Xen virtual machine take Cambridge University is as representative, and its only needs to revise part operation system code and can realize virtual.
At present, most of virtual machines provide multiple Starting mode, both can from the virtual image file start, also can from network startup, can also start from a file directory.Cloud computing offers the user and uses take virtual machine as unit of account, provide service in order to give simultaneously numerous users, needs the virtual machine of magnanimity.The magnanimity virtual machine need to start from the magnanimity root file system, because a virtual machine must corresponding root file system.In order to store the file of magnanimity, at present popular is distributed file system.But traditional distributed file system needs that all one deck file system nodes is arranged, the data inflow and outflow of all memory nodes all must be through this intermediate node, and this extra intermediate node becomes the bottleneck of whole file system when causing a plurality of clients to carry out Concurrency Access.The bottleneck problem of this in order to solve " single service node " has produced the new architecture that a kind of metadata and file data are stored respectively in distributed file system.Client can be obtained by different links attribute information and the document entity content of file, and the file reading entity content that can walk abreast, to improve the concurrent handling capacity of whole system.Present popular distributed file system GFS and HDFS, the design of just adopting meta data server to separate with data server, but all only comprise a meta data server and (in the GFS file system, be called Master, in HDFS, be called NameNode) to simplify metadata management, and metadata information all is stored in the internal memory of meta data server to accelerate access speed, and the memory headroom of meta data server obviously can become the bottleneck of file system extensibility like this.
The Xen virtual machine is the current the most popular software virtual machine of increasing income, and the root file system of virtual machine can be installed on the pNFS server, provides from the method for pNFS client terminal start-up virtual machine.This starting method is very useful in Xen virtual machine (vm) migration function, when carrying out virtual machine (vm) migration, need this virtual machine root file system from the sourcesink host migration to the purpose host, and pNFS just in time provides the method that file system is mounted to simultaneously two hosts, so only needs can finish by the path that changes root file system the migration of virtual machine file system.
PNFS also is the design of adopting meta data server to separate with data server, and can allow client concurrently direct and data server carry out reading and writing data.The pNFS distributed file system is by a meta data server (Metadata Server, MDS) and a plurality of data server (Data Server, DS) form, MDS is in charge of file system title and the distributed intelligence of file data on DS, and DS is responsible for storing data.Virtual machine provides the basic calculating unit of service in virtual for cloud computing.Separate by controlling with data stream, pNFS allows client directly and memory node carries out alternately.PNFS provides many attracting characteristics.At first, it supports various layout types (block, file, object), can be based upon on existing any I/O and the storage protocol.Secondly, it provides easily migration and the upgrade method of row, to improve system performance and scalability.Again, at present numerous distributed file systems, pNFS is few one of project of increasing income.Therefore, the pNFS range of application is very extensive, from the data center to the workstation, high-speed servers etc.
Fig. 1 has shown in existing pNFS system, the polymerization readwrite performance that large file and small documents increase along with number of clients.As can be seen from Figure 1, for large file, along with the number clients purpose increases, the polymerization readwrite performance is linear growth; Yet the polymerization speed of small documents has just reached peak value when two or three clients.When pNFS storage small documents, the size of file is less than minute block size of pNFS storage file, and all file datas will be stored on the data server, and other data server all is empty.In this case, in fact the pNFS server deteriorates to the NFS that only has a data server, can't carry out the concurrent reading and concurrent writing of data.Learn by statistics, most of (SuSE) Linux OS files are smaller file (in Ubuntu Server11.04,96% operating system file is less than 200KB), this moment, dummy machine system operated on the pNFS file system, if the concurrent startup of magnanimity virtual machine moves, can have a strong impact on the performance of virtual machine.
Summary of the invention
(1) technical matters to be solved
The object of the present invention is to provide a kind of pNFS system high efficiency to support the magnanimity virtual machine to start online the method for operation, meta data server pressure is excessive when having the mass file concurrent reading and writing that exists in the pNFS distributed file system now to solve, and file data skewness weighing apparatus on each data server, thereby can't the efficient online problem that starts operation of good support magnanimity virtual machine.
(2) technical scheme
In order to solve the problems of the technologies described above, the present invention proposes a kind of pNFS system high efficiency and support the magnanimity virtual machine to start online the method for operation, said method comprising the steps of:
S1, client are obtained the layout types information of pNFS system from meta data server, and the layout types of judging described pNFS system whether with self compatibility, if compatible, then enter step S2,
Wherein, described pNFS system comprises several data servers;
S2, described client are to the layout information of the required file of described meta data server request, and described meta data server instant computing goes out the layout information of described file and sends to described client;
S3, described client be according to the layout information of described file, to the solid data of the described file of data server parallel access of correspondence.
Optionally, further comprise step after the step S3:
After S4, described client are carried out write operation to described file, submit the layout modification information of described file to described meta data server.
Optionally, described layout information comprises the distributed intelligence in data server of data server device id and file,
Wherein, described data server device id is the unique identifying number that described meta data server is distributed to each data server.
Optionally, the distributed intelligence of described file in data server uses file handle to identify.
Optionally, among the step S2, the layout information that described meta data server calculates described file specifically comprises:
Take the index node inode of described file as condition code, carry out Hash calculation, and further obtain the data server device id of described file.
Optionally, step S2 further comprises:
Utilize the data server device id of described file, generate the file handle sequence of corresponding data server.
Optionally, step S3 specifically comprises:
Described client is according to the data server device id of described file, obtain IP address and the port numbers of corresponding data server to described meta data server, and according to IP address and the port numbers of described data server, to the described file of described data server access.
(3) beneficial effect
Compare with existing pNFS system, the technical scheme that the present invention proposes has following advantage:
1, simplifies MDS end member data technique, alleviated when mass file is concurrent to be read the pressure to MDS.The technical scheme that the present invention proposes is simplified the metadata information of the upper storage of MDS, only kept the distributed intelligence of file in storage system, and be different from traditional method that information is stored in the internal memory, but when this information of client query, carry out instant computing.
2, can dynamic equalization DS data distribute, make file data be distributed to more equably each DS.For different files, as condition code, carry out the Hash operation with the inode of file, calculate different DS distribution situations, thereby reach the purpose of load balancing.
The technical scheme that the present invention proposes is particularly useful for the online start-up and operation of magnanimity virtual machine under the general scene, can greatly improve the pNFS distributed file system starts operation to the magnanimity virtual machine network tenability, significantly be lifted at the readwrite performance of virtual machine file system in the high concurrent situation.
Description of drawings
Fig. 1 is the polymerization readwrite performance schematic diagram of pNFS system in the prior art.
Fig. 2 is the system architecture schematic diagram of the technical scheme that proposes of the present invention.
Fig. 3 is the distribution contrast schematic diagram of one embodiment of the present of invention File on data server
Fig. 4 contrasts schematic diagram the start-up time of DomU virtual machine in one embodiment of the present of invention.
Embodiment
The present invention proposes a kind of pNFS system high efficiency and support the magnanimity virtual machine to start online the method for operation, said method comprising the steps of:
S1, client are obtained the layout types information of pNFS system from meta data server, and the layout types of judging described pNFS system whether with self compatibility, if compatible, then enter step S2,
Wherein, described pNFS system comprises several data servers;
S2, described client are to the layout information of the required file of described meta data server request, and described meta data server instant computing goes out the layout information of described file and sends to described client;
S3, described client be according to the layout information of described file, to the solid data of the described file of data server parallel access of correspondence.
Preferably, further comprise step after the step S3:
After S4, described client are carried out write operation to described file, submit the layout modification information of described file to described meta data server.
Below in conjunction with drawings and Examples, the specific embodiment of the present invention is described in further detail.
The present invention is on the basis of pNFS distributed file system, the characteristic of separating with file data for its metadata, and the VME operating system file mostly is greatly the characteristic of small documents, proposed: (1) simplifies Meta Server end member data technique, alleviates when the magnanimity operating system file is concurrent to be read the pressure to Meta Server; (2) dynamic equalization Data Server data distribution technique makes file data be distributed to more equably each Data Server.
The pNFS distributed file system is divided into Block, File, three kinds of distribution forms of Object, in a kind of preferred implementation of the present invention, can select the file system based on the File layout.As shown in Figure 2, pNFS distributed file system structure mainly comprises meta data server (MDS), data server (DS) and pNFS client, and wherein the pNFS client is the host of virtual machine operation.Client is by obtaining the distribution situation of file data on DS to the inquiry of MDS end, and then direct and DS memory node carries out the concurrent reading and concurrent writing of file data.For the mass file storage system, often require concurrency very high, this time the excessive access of MDS is tended to become the bottleneck of system.In order to alleviate this problem, the present invention simplifies the metadata information of the upper storage of MDS, only keep the distributed intelligence of file on memory device, and be different from traditional method that information is stored in the internal memory, but when this information of client query, carried out instant computing.
As shown in Figure 2, when client was carried out reading and writing of files in this storage system, its flow process was as follows:
S1, client send the GETATTR operation requests to meta data server, obtain the layout types information of current storage system, and the layout types that only has meta data server to return is that client is supported, just can carry out follow-up operation.
S2, client need the layout information of reading and writing of files to the meta data server request, this layout information comprises the distributed intelligence on data server of data server device id and file.Wherein the data server device id is corresponding to one group of data server, the order of this data server has also been indicated the read-write order of file data at each DS, that is to say, the first data block of file is distributed in DS1, second data block is distributed on the DS2, the like.For file data is evenly distributed on each DS, different files uses different DS sequences, and the DS sequence uses following formula to generate:
DeviceID_Locator=((Hash(file_inode)+i)mod?DataServers_Length)
Wherein, file_inode is the index node inode information (inode of each file is unique) of this document, and i represents i the data block of this document.When carrying out Hash calculation, the sign of data block is not placed on the distribution that can guarantee in the hash function for large file evenly distributes yet.If use Hash (file_inode+i), experimental results show that when data server increases, can cause the maximum server of memory space and the minimum server difference of storage obviously to increase, thereby cause load imbalance.
The distributed intelligence of file on data server uses file handle to identify, and different handle sequences are corresponding to different data servers, and the file handle sequence uses following formula to generate:
FILEHANDLES[i]=GET_FH(file_path[(Hash(file_inode)+i)modDataServers_Length])
The data server device id that S3, client are obtained by step S2 obtains the details of file data memory device to meta data server.The information such as the memory device IP that client need to read by this acquisition of information, port numbers.The file handle information of storing device information in must corresponding step S2 can correctly be read and write to guarantee client.For distinct device ID, the memory node information exchange is crossed following formula and is generated:
DSLIST[i]＝GET_DSADDR(DataServers[(DeviceID+i)%N])
After client gets access to storing device information and file handle information, just can be directly and the file data read-write operation that walks abreast of memory device (being data server).
If S4 this document is operating as write operation, modification situation, the actual size of updating file and the end of file situation of the layout information of distribution before then client need to be submitted to meta data server.
In order to verify that method that the present invention proposes starts the support effect of operation online to virtual machine, the experiment that we have designed concurrent start-up time of virtual machine comes and compare the start-up time of traditional pNFS storage.
In the experiment, at first the operating system root file with Ubuntu Server11.04 is stored in respectively on the storage system (tpNFS) of traditional pNFS and the present invention's proposition.As shown in Figure 3, pNFS represents traditional pNFS among the figure, tpNFS represents to adopt the storage system of the file access method that the present invention proposes, when root file system is stored on traditional pNFS, file data is concentrated and is distributed on 5 the data servers in front, the file that distributes on the data server of back is (it is top that the file less than 3% has data to be distributed in) seldom, is empty substantially; In the time of on root file system being stored in the storage system that the present invention proposes, the data on each data server distribute more even, and all files are evenly distributed on the data server basically.Then, virtual machine start-up time on two storage systems respectively of test varying number.As shown in Figure 4, the file access method that the present invention proposes under different virtual machine quantity is all fast than the reading/writing method of traditional pNFS, and along with the increase of virtual machine quantity (from 2 DomU virtual machines to 32 DomU virtual machine, each host starts 2 DomU virtual machines), the effect of the technical scheme that the present invention proposes is than good the getting more and more of traditional pNFS, when virtual machine quantity was 32, the virtual machine activation time decreased that adopts technical scheme of the present invention was 46.97% of traditional pNFS upper start-up time.Can find out from experimental result, file access method and system that the present invention proposes provide good storage support for the virtual machine root file system, efficiently supported virtual machine network to start operation, particularly in the situation that a plurality of virtual machine is high concurrent, its advantage is more obvious.
The above only is preferred implementation of the present invention; should be pointed out that for the person of ordinary skill of the art, under the prerequisite that does not break away from the technology of the present invention principle; can also make some improvement and replacement, these improvement and replacement also should be considered as protection scope of the present invention.
Claims (7)
1. a pNFS system high efficiency supports the magnanimity virtual machine to start online the method for operation, it is characterized in that, said method comprising the steps of:
S1, client are obtained the layout types information of pNFS system from meta data server, and the layout types of judging described pNFS system whether with self compatibility, if compatible, then enter step S2,
Wherein, described pNFS system comprises several data servers;
S2, described client are to the layout information of the required file of described meta data server request, and described meta data server instant computing goes out the layout information of described file and sends to described client;
S3, described client be according to the layout information of described file, to the solid data of the described file of data server parallel access of correspondence.
2. pNFS system high efficiency according to claim 1 supports the magnanimity virtual machine to start online the method for operation, it is characterized in that, further comprises step after the step S3:
After S4, described client are carried out write operation to described file, submit the layout modification information of described file to described meta data server.
3. pNFS system high efficiency according to claim 1 and 2 supports the magnanimity virtual machine to start online the method for operation, it is characterized in that, described layout information comprises the distributed intelligence in data server of data server device id and file,
Wherein, described data server device id is the unique identifying number that described meta data server is distributed to each data server.
4. pNFS system high efficiency according to claim 3 supports the magnanimity virtual machine to start online the method for operation, it is characterized in that, the distributed intelligence of described file in data server uses file handle to identify.
5. pNFS system high efficiency according to claim 4 supports the magnanimity virtual machine to start online the method for operation, it is characterized in that, among the step S2, the layout information that described meta data server calculates described file specifically comprises:
Take the index node inode of described file as condition code, carry out Hash calculation, and further obtain the data server device id of described file.
6. pNFS system high efficiency according to claim 5 supports the magnanimity virtual machine to start online the method for operation, it is characterized in that, step S2 further comprises:
Utilize the data server device id of described file, generate the file handle sequence of corresponding data server.
7. pNFS system high efficiency according to claim 3 supports the magnanimity virtual machine to start online the method for operation, it is characterized in that, step S3 specifically comprises:
Described client is according to the data server device id of described file, obtain IP address and the port numbers of corresponding data server to described meta data server, and according to IP address and the port numbers of described data server, to the described file of described data server access.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201210592629.0A CN103077055B (en) | 2012-12-31 | 2012-12-31 | PNFS system supports the method for the online startup optimization of magnanimity virtual machine |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201210592629.0A CN103077055B (en) | 2012-12-31 | 2012-12-31 | PNFS system supports the method for the online startup optimization of magnanimity virtual machine |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN103077055A true CN103077055A (en) | 2013-05-01 |
| CN103077055B CN103077055B (en) | 2016-01-13 |
Family
ID=48153592
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201210592629.0A Active CN103077055B (en) | 2012-12-31 | 2012-12-31 | PNFS system supports the method for the online startup optimization of magnanimity virtual machine |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN103077055B (en) |
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106506587A (en) * | 2016-09-23 | 2017-03-15 | 中国人民解放军国防科学技术大学 | A kind of Docker image download methods based on distributed storage |
| CN107678836A (en) * | 2017-10-12 | 2018-02-09 | 新华三大数据技术有限公司 | The test data acquisition methods and device of cluster |
| CN111399764A (en) * | 2019-12-25 | 2020-07-10 | 杭州海康威视系统技术有限公司 | Data storage method, data reading device, data storage equipment and data storage medium |
| CN114116655A (en) * | 2021-12-08 | 2022-03-01 | 北京交研智慧科技有限公司 | Parallel file storage method, device, equipment and storage medium based on virtual machine |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2010077972A2 (en) * | 2008-12-16 | 2010-07-08 | Netapp, Inc. | Method and apparatus to implement a hierarchical cache system with pnfs |
| CN101996217A (en) * | 2009-08-24 | 2011-03-30 | 华为技术有限公司 | Method for storing data and memory device thereof |
| CN102164161A (en) * | 2011-01-10 | 2011-08-24 | 清华大学 | Method and device for performing file layout extraction on parallel file system |
-
2012
- 2012-12-31 CN CN201210592629.0A patent/CN103077055B/en active Active
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2010077972A2 (en) * | 2008-12-16 | 2010-07-08 | Netapp, Inc. | Method and apparatus to implement a hierarchical cache system with pnfs |
| CN101996217A (en) * | 2009-08-24 | 2011-03-30 | 华为技术有限公司 | Method for storing data and memory device thereof |
| CN102164161A (en) * | 2011-01-10 | 2011-08-24 | 清华大学 | Method and device for performing file layout extraction on parallel file system |
Non-Patent Citations (2)
| Title |
|---|
| 李洪昌 等；: ""一种基于pNFS协议的对象文件系统的设计与实现"", 《小型微型计算机系统》 * |
| 李洪昌等；: ""一种基于pNFS协议的对象文件系统的设计与实现"", 《小型微型计算机系统》, vol. 33, no. 11, 30 November 2012 (2012-11-30) * |
Cited By (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106506587A (en) * | 2016-09-23 | 2017-03-15 | 中国人民解放军国防科学技术大学 | A kind of Docker image download methods based on distributed storage |
| CN106506587B (en) * | 2016-09-23 | 2021-08-06 | 中国人民解放军国防科学技术大学 | Docker mirror image downloading method based on distributed storage |
| CN107678836A (en) * | 2017-10-12 | 2018-02-09 | 新华三大数据技术有限公司 | The test data acquisition methods and device of cluster |
| CN107678836B (en) * | 2017-10-12 | 2021-09-03 | 新华三大数据技术有限公司 | Cluster test data acquisition method and device |
| CN111399764A (en) * | 2019-12-25 | 2020-07-10 | 杭州海康威视系统技术有限公司 | Data storage method, data reading device, data storage equipment and data storage medium |
| CN114116655A (en) * | 2021-12-08 | 2022-03-01 | 北京交研智慧科技有限公司 | Parallel file storage method, device, equipment and storage medium based on virtual machine |
Also Published As
| Publication number | Publication date |
|---|---|
| CN103077055B (en) | 2016-01-13 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN108885582B (en) | Multi-tenant memory services for memory pool architecture | |
| CN102521063B (en) | Shared storage method suitable for migration and fault tolerance of virtual machine | |
| US8972986B2 (en) | Locality-aware resource allocation for cloud computing | |
| US20210141917A1 (en) | Low latency access to physical storage locations by implementing multiple levels of metadata | |
| US10990441B2 (en) | Multi-level job processing queues | |
| CN103310000B (en) | Metadata management method | |
| US20210216210A1 (en) | Optimized migration of data between file systems of a storage array | |
| US10824369B2 (en) | Elastic method of remote direct memory access memory advertisement | |
| CN103077055B (en) | PNFS system supports the method for the online startup optimization of magnanimity virtual machine | |
| US9110820B1 (en) | Hybrid data storage system in an HPC exascale environment | |
| US11080207B2 (en) | Caching framework for big-data engines in the cloud | |
| CN104407879A (en) | A power grid timing sequence large data parallel loading method | |
| CN109933312A (en) | A method of containerization relevant database I/O consumption is effectively reduced | |
| CN103491155A (en) | Cloud computing method and system for achieving mobile computing and obtaining mobile data | |
| CN103595799A (en) | Method for achieving distributed shared data bank | |
| CN102420850B (en) | Resource scheduling method and system thereof | |
| CN103049546B (en) | The method and apparatus of a kind of management, access system daily record | |
| Nguyen et al. | An index scheme for similarity search on cloud computing using mapreduce over docker container | |
| CN101783814A (en) | Metadata storing method for mass storage system | |
| Li et al. | Dynamic hashing: Adaptive metadata management for petabyte-scale file systems | |
| Zhu et al. | Marlin: Taming the big streaming data in large scale video similarity search | |
| Jin et al. | Optimization of task assignment strategy for map-reduce | |
| US20200125650A1 (en) | Lazy virtual filesystem instantiation and caching | |
| Bao | Performance evaluation for traditional virtual machine placement algorithms in the cloud | |
| Priya Ponnuswamy et al. | File retrieval and storage in the open source cloud tool using digital bipartite and digit compact prefix indexing method |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| C14 | Grant of patent or utility model | ||
| GR01 | Patent grant |