CN106874383A - A kind of decoupling location mode of metadata of distributed type file system - Google Patents
A kind of decoupling location mode of metadata of distributed type file system Download PDFInfo
- Publication number
- CN106874383A CN106874383A CN201710016284.7A CN201710016284A CN106874383A CN 106874383 A CN106874383 A CN 106874383A CN 201710016284 A CN201710016284 A CN 201710016284A CN 106874383 A CN106874383 A CN 106874383A
- Authority
- CN
- China
- Prior art keywords
- metadata
- file
- catalogue
- directory
- node
- 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
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/10—File systems; File servers
- G06F16/13—File access structures, e.g. distributed indices
- G06F16/134—Distributed indices
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/10—File systems; File servers
- G06F16/18—File system types
- G06F16/182—Distributed file systems
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Data Mining & Analysis (AREA)
- Databases & Information Systems (AREA)
- Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
Abstract
The invention discloses a kind of decoupling location mode of metadata of distributed type file system, including:Metadata to distributed file system is separated, to obtain the metadata of the metadata of catalogue, the metadata of directory entry and file;The directory metadata is centrally stored in directory metadata index node, and not comprising the pointer of sensing directory entry.Directory operation is performed according to the directory inode.Associated file metadata is stored in same node after each directory entry metadata is split, and sets up the reverse indexing for pointing to directory metadata.The invention has the advantages that:Reduce the information exchange between each node when distributed file system accesses metadata, reduce the delay of metadata access, simultaneously, method by separating directory content, decoupling High relevancy between file and catalogue, handling capacity very high can be reached, so as to improve treatment effeciency of the distributed file system for metadata.
Description
Technical field
The present invention relates to computer realm, and in particular to a kind of decoupling distribution side of metadata of distributed type file system
Method.
Background technology
Distributed file system is a kind of Novel storage system for supporting mass data storage, is widely used in data
The heart, Chao Suan centers and publicly-owned cloud platform.Distributed file system has the advantages that many good relative to traditional centralised storage.
Data storage can such as be carried out extending transversely, the capacity of storage can be dynamically expanded by way of increasing memory node,
And ensure the Synchronous lifting of access throughput.Secondly, distributed file system has flexible relative to traditional centralised storage
Fault-tolerant strategy, it is possible to use copy mechanism and correcting and eleting codes carry out distributed fault-tolerant.Distributed file system can also be used
More cheap storage and computing device goes to build a large-scale storage cluster, to ensure the access of mass data.But
The access standard (POSIX) of file system is limited to, the metadata access of distributed file system often becomes its performance
Bottleneck.The access of its metadata cannot often meet the demand of high-throughput and low delay, but in actual system, exceed
The data access of more than half is needed by metadata node.In order to solve the scalability of metadata of distributed type file system,
Existing technology mainly has following three:
The characteristics of a kind of distributed meta data point spread method for being to be based on Dictionary tree, this method is to be distributed
The name space of formula file system is divided into different subtrees according to subdirectory, and each subtree independence is stored in some node,
And the node of the dynamic regulation storage of load for more accessing.The advantage of this mode is can be according to the synchronous dynamic of load
The position that accesses of adjustment, but this mode cannot solve the problems, such as the path backtracking of file access, when accessing file
When, it is necessary to access all catalogues in whole path, and these catalogues are not often stored in same node, often result in
Larger access delay.
Another kind is the metadata profile method based on hash algorithm, is characterized in the file in a catalogue by breathing out
Uncommon mode is by data allocations to different nodes.The advantage of this mode has substantial amounts of file in being directed to a catalogue
When, the load of file access can be reduced.But the scaling concern of catalogue cannot be solved.
The third method is the method by using key value database storage file metadata, and this method make use of key assignments
Database access is fast, and the characteristics of time delay is low, but this method still has the path searching of path such as first method presence
Problem, cannot still solve the problems, such as relatively low during access time delay.
In order to solve the problems, such as path delay, these methods are often in client-cache metadata, but this brings again
The expense of many inconsistencies, so that cannot be from more this upper solve problem.
The content of the invention
It is contemplated that at least solving one of above-mentioned technical problem.
Therefore, the decoupling distribution side it is an object of the present invention to propose a kind of metadata of distributed type file system
Method, to solve the metadata profile of distributed file system, throughput is not high and postpones relatively low problem.
To achieve these goals, embodiment of the invention discloses that a kind of metadata of distributed type file system it is decoupling
Location mode, comprises the following steps:S1:Metadata to distributed file system is separated, to obtain directory inode
Metadata, the metadata of directory entry and file metadata;S2:The metadata of the catalogue is arranged on directory index section
Point;S3:Each directory entry is split according to the distribution situation of file, and it is associated in the node storage of file storage
Directory entry, and set up the reverse indexing for pointing to directory metadata.
Further, the directory operation includes establishment, the deletion of catalogue, reading catalogue, the institute of acquisition catalogue of catalogue
The user belonging to user's group and change catalogue where having metadata, change catalogue.
Further, also include:The globally unique mark for determining file is provided;The overall situation of the file accessed required for calculating
The cryptographic Hash of the mark;The node that metadata is deposited is positioned according to the cryptographic Hash.
Further, the fullpath for being designated file.
Further, also include:When establishment file or catalogue, one is created in the node of establishment file or catalogue
All directory entries in the parent directory path comprising the file or the catalogue;If the directory entry is all or part of
Created in the node, then create remaining directory entry.
Further, also include:When a file is deleted, the metadata of node and institute where the file are deleted
State the project that the corresponding directory entry metadata of node where file points to the file.
Further, also include:When be read out catalogue or deltree operation when, access all of metadata node,
To obtain all directory entries under reading catalogue or deltreeing.
Further, also include:Client-cache is provided, wherein, the directory metadata of the client-cache is used for visitor
Determine whether the authority with establishment file when the establishment file of family end;The client access file metadata when,
The metadata of catalogue is accessed, to obtain access rights;When the client has access rights, the metadata of file is accessed.
Further, also include:In the cache invalidation of described directory metadata client, directory metadata is carried out
The change of the authority of catalogue and the deletion for catalogue.
The decoupling location mode of metadata of distributed type file system according to embodiments of the present invention, it is all of for file
Metadata operation at most access 2 minor nodes, under the caching situation of directory metadata, it is only necessary to access a minor node, it prolongs
When be only once access round RTT, due to being stored using key assignments, thus for metadata obtain time delay it is very low,
Be can be ignored in the RTT time delays of Ethernet, institute's distributed file system that can in this way be effectively reduced is accessed
Information exchange during metadata between each node, reduces the delay of metadata access, meanwhile, by separating directory content
Method, decoupling High relevancy between file and catalogue, can reach handling capacity very high, so as to improve distributed text
Treatment effeciency of the part system for metadata.
Additional aspect of the invention and advantage will be set forth in part in the description, and will partly become from the following description
Obtain substantially, or recognized by practice of the invention.
Brief description of the drawings
Of the invention above-mentioned and/or additional aspect and advantage will become from description of the accompanying drawings below to embodiment is combined
Substantially and be readily appreciated that, wherein:
Fig. 1 is the flow chart of the decoupling location mode of the metadata of distributed type file system of the embodiment of the present invention;
Fig. 2 is the overall system architecture figure of one embodiment of the invention;
Fig. 3 is the uncoupled schematic diagram of contents segmentation of one embodiment of the invention.
Specific embodiment
Embodiments of the invention are described below in detail, the example of the embodiment is shown in the drawings, wherein from start to finish
Same or similar label represents same or similar element or the element with same or like function.Below with reference to attached
It is exemplary to scheme the embodiment of description, is only used for explaining the present invention, and is not considered as limiting the invention.
With reference to following description and accompanying drawing, it will be clear that these and other aspects of embodiments of the invention.In these descriptions
In accompanying drawing, specifically disclose some particular implementations in embodiments of the invention to represent implementation implementation of the invention
Some modes of the principle of example, but it is to be understood that the scope of embodiments of the invention is not limited.Conversely, of the invention
Embodiment includes all changes, modification and the equivalent that fall into the range of the spiritual and intension of attached claims.
Below in conjunction with the Description of Drawings present invention.
Fig. 1 is the flow chart of the decoupling location mode of the metadata of distributed type file system of the embodiment of the present invention.Such as Fig. 1
Shown, the decoupling location mode of metadata of distributed type file system according to embodiments of the present invention is comprised the following steps:
S1:Metadata to distributed file system is separated, to obtain metadata, the metadata of directory entry of catalogue
With the metadata of file.
S2:The metadata of the catalogue is arranged on directory inode.
S3:Each directory entry is split according to the distribution situation of file, and the node deposited in file is stored therewith
Related directory entry, and set up the reverse indexing for pointing to directory metadata.
In one embodiment of the invention, the directory metadata of file system is centrally stored in a node.At this
Under the mode of kind, the metadata information of directory inode is not comprising the address for pointing to directory entry metadata.Only retain basic
The creation time of catalogue data, including but not limited to catalogue, the capability identification of catalogue, the group where catalogue is identified, belonging to catalogue
ID.Herein on basis, for most of related to the metadata of directory inode with directory entry metadata
Unrelated metadata operation all will be carried out only on this node of storage catalogue index metadata.Directory operation includes catalogue
Establishment, the deletion of catalogue, read catalogue, obtain catalogue all metadata, change catalogue where user's group and change mesh
User belonging to record.
In one embodiment of the invention, also machine is stored including a kind of distributed file metadata based on Hash
System.This memory mechanism is supported for the storage and access for file metadata to expand to multiple nodes, so as to reach balance system
The purpose of system load.This algorithm is used can be in one mark of file of globally unique determination:When client is carried out to file
During metadata operation, the cryptographic Hash of the globally unique identifier of the file that client is accessed required for by calculating, positioning file institute
The node of storage, operates in the node to metadata.This method ensure that for All Files metadata operation all
An information for metadata node will at most be changed.
In one embodiment of the invention, it is designated the fullpath of file.
In one embodiment of the invention, the also reverse storage method including a kind of directory entry.This method passes through will
The metadata of directory entry is assigned to multiple nodes after being split, it is ensured that need not increase during the establishment and deletion of file
Plus extra node visit expense.Its distribution method is when establishment file or catalogue, it is not necessary to which modification is literary except creating
The metadata information on node outside part node, but the node establishment one in establishment file or catalogue includes this document
Or all directory entries in the parent directory path of catalogue.If all or part of of these directory entries is created in the node
Build, then create remaining directory entry.This directory entry ensure that for when establishment and deletion text by this reverse storage method
During part, the modification to multiple metadata nodes is not resulted in, reduce and access the distributed synchronization expense that multiple nodes bring.
In one embodiment of the invention, when a file is deleted, the metadata of node where file is deleted
Directory entry metadata corresponding with node where file points to the project of file.
In one embodiment of the invention, also include:When be read out catalogue or deltree operation when, access all
Metadata node, read catalogue or all directory entries under deltreeing to obtain.
In one embodiment of the invention, also include:Client-cache is provided, wherein, the catalogue unit of client-cache
Data are used to determine whether the authority with establishment file when client establishment file;Client is accessing first number of file
According to when, access catalogue metadata, to obtain access rights;When client has access rights, the metadata of file is accessed.
In one embodiment of the invention, also include:In the cache invalidation of described directory metadata client, enter
The change of the authority of the catalogue of column catalogue metadata and the deletion for catalogue.
To make it is further understood that the present invention, will be described in detail by following examples.
As shown in Figure 2, there is provided a kind of metadata node of storage catalogue, this metadata node is used to process distribution
All requests on directory metadata in file system.In implementation process, it is led to using the RPC agreements based on Ethernet
The POSIX access interfaces for crossing offer standard receive the request from client.It is divided into four modules, and one is to provide and performs unit
The access interface of data request operation, one is key assignments thesaurus, for metadata to be persisted into disk.In addition, also
The module for having a storing directory content caches the directory content of each node.When metadata node is started, it can scan key
Each value in value storage builds the content of catalogue.
In the treatment for directory metadata, the metadata of catalogue includes, the creation time of catalogue, the authority mark of catalogue
Know, the group mark where catalogue, the ID belonging to catalogue, the globally unique identifier of catalogue.Wherein, the path length of catalogue
It is an elongated character string, the metadata of remaining every catalogue is respectively a fixed length mark of 8 bytes.This first number
According to node support the associative operation on catalogue, including catalogue establishment, the deletion of catalogue, read catalogue, obtain catalogue institute
There is metadata, the user's group where changing catalogue changes the user belonging to catalogue.
Wherein the establishment process of catalogue includes being received from client the request for createing directory, and this request includes client
The path for createing directory, the permission mode that client creaties directory, user's group mark and ID that client creaties directory.
Wherein the permission mode of catalogue includes the access limit of the owner of catalogue, the access limit and other users of user's group user
Access limit.After node side receives these requests, the parent directory that can be creatied directory to client first carries out authority inspection
Look into and confirm, when establishment condition is determined for compliance with, by the storage of the metadata write-in directory metadata node of catalogue.Wherein accord with
The establishment condition of conjunction includes the parent directory formulated in the presence of it, and possesses the access rights of parent directory.
For the deletion of catalogue, all metadata of catalogue are read, the user's group and change catalogue where changing catalogue
Whether these are operated attribute, after the request of client is received, will all be first checked for catalogue and be whether there is and possess access right
Limit.When it is determined that after this two, just being operated accordingly to catalogue.Compare special, when deleting catalogue,
Must determine that the file of its subdirectory and subdirectory is deleted.
In whole framework, directory metadata node only has 1 in logic in the entire system.First number of all of catalogue
According to being all stored in a metadata node.In order to ensure its reliability, it is possible to use the method for backed up in synchronization and can not use
The method of distribution extension.
Fig. 2 also describes a kind of node of storage file metadata, and this metadata node is used to process distributed document
The metadata request of all about file of system, and distributed storage file system metadata.It is right that includes supporting
In the standard POSIX access interfaces of file operation, the metadata management layer of classification storage, and a key are carried out for metadata
Metadata is persisted to disk by value storage.
The metadata of each file include first aspect described in its parent directory GUID, file name,
The access time of file, the authority of file, the user's group mark where file, the modification time of file, the access of file content
Time, the size of file, the size of blocks of files creates the overall identification of the metadata node of this document, the place of establishment file
Metadata node.The wherein GUID of its parent directory and the title of filename is combined, and composition one is elongated
Character string.Other metadata fields are then the field of the byte of fixed length 8.This metadata supports the phase for file metadata
Close operation, including file establishment, the deletion of file obtains the metadata of file, changes the user's group of file, change file institute
The user of category, changes the authority of file, and the read-write of file changes the size of file.
Wherein, the establishment process of file is divided into two stages, first stage, and client needs to access establishment file institute first
The catalogue of category, client needs to access the directory metadata node described in first aspect, it is determined whether with the wound under the catalogue
Build the authority of file.After it is determined that being a legal request to create, into second stage, client will transmit a request to text
Part metadata node, the character string that file node will be combined with the title of the globally unique identifier of parent directory and filename
Go to be indexed on file metadata node, when it is determined that there is no this document on the node, then establishment file success.
For the metadata operation of other files, then firstly the need of file is found, determine whether afterwards with access right
Limit, if possessing access rights, is changed it accordingly.In these modifications, the modification for file read-write is necessary
To wait until that the data of file have been written into the storage system of rear end or successfully read ability from the storage system of rear end
The modification of data is enough completed, this process needs to be defined as a process for affairs.
Metadata node is a kind of network node based on RPC communication, needs to open corresponding clothes when initialization
Business port, while starting the database of storage catalogue metadata.For the treatment that metadata operation is asked, the side of multithreading is used
Formula is concurrent to be processed request.
As shown in Fig. 2 file metadata node is only multiple in logic in the entire system.The metadata of all of file
The character string that is made up of the unique identifier and filename that calculate its parent directory determines the position of the node that file is deposited.
This computational methods are ensured using the hash algorithm of uniformity.This method can ensure that it possesses preferable autgmentability, while
The metadata node of dynamic extended file can also be supported.
As shown in Fig. 2 its framework also includes a kind of distributed type file system client side with caching.In this client
On, application program can access distributed storage by directly accessing the storehouse of distributed file system offer.In linux system
In, user can be just supported in client directly by file after can calling this storehouse by User space file system (FUSE)
System is mounted on disk.After client terminal start-up, a Directory caching can be set up in client, this caching is placed on accessing
On storehouse.
Client builds different access strategies by RPC agreements when accessing for different access process.For mesh
The metadata operation of record, including mkdir creaties directory, and getattr obtains an attribute for file, catalogue or file (here
Only obtain the metadata of catalogue), chmod file/directory priority assignation orders, chown is Multi-User Multi-Task operating system, institute
Some files all have its owner (Owner).Client will directly access directory metadata node, for different operations, visitor
Family end can provide different parameters.Node side is received after directly being processed after asking and returns to client.For the metadata of file
Operation, including open, read, write, getattr, truncate etc., it checks his father's mesh firstly the need of Directory caching is accessed
Whether record has access rights, if not having the information of correlation in Directory caching, client will first access directory metadata section
Point, by the Directory caching of the metadata cache of parent directory to client, checks authority afterwards, it is determined that after possessing authority, client
Metadata operation is initiated in end with file metadata node again asks.What needs were particularly pointed out, (read in catalogue for readdir
Hold) this operation, client will initiate All Files metadata node and directory metadata node to read directory content
The directory content of respective storage is returned to client by request, these nodes, and client is by these Content Organizings into ordered list
Return to user.For this operation of rmdir (deltreeing), client is needed to all of file metadata node and catalogue
Metadata node initiate obtain directory content request, when all returns request be space-time, then confirm be for the deletion of catalogue
Legal, now could initiate the request that deltrees to file metadata node.
Caching on client, the example also needs to process following situation.Due to there are some visitors in the process of running
The content of catalogue is changed at family end, so the directory metadata for caching needs the out-of-service time of setting caching, for catalogue unit
The change of the authority of the catalogue of data and need to wait the cache invalidation of all directory metadata clients for the deletion of catalogue
Operation can be continued.So when certain client sends both requests, request will not be performed, but be waited until at once
After the cache invalidation time time-out of directory metadata node sets, can just enter the renewal of column catalogue and delete work.
Client as shown in Figure 2 also needs to a configuration file to deposit the map of the overall situation.Global map includes mesh
IP address where record metadata node and file metadata node, the globally unique numbering of each file metadata node, base
In this global map and numbering, client just can calculate the position for determining file node by hash algorithm, while can
Enter row information by network and corresponding metadata node to exchange.During client initialization, client can be by the overall situation
Map reads in, and directory metadata node and file metadata node set up heartbeat link, and each will determine catalogue for a period of time
Whether also normally worked with file metadata node.
As shown in figure 3, it is necessary to carry out cutting to the metadata of file in the organizational process of metadata, in this process
In, 4 kinds of metadata structures of this example definition, including catalogue metadata (d-inode), the content (d-entry) of catalogue, text
The metadata (f-inode) of part, the content (f-content) of file.In traditional file system, d-inode passes through d-
Entry can index f-inode, and f-inode can index the particular content of file.And in this example, it is proposed that one
Plant new metadata organization method.This metadata cutting method includes each Nodes Self-organized directory content of oneself.Such as Fig. 2
It is shown, by decoupling, the d-entry of script is assigned to each d-inode.Thus d-entry is solved to file and mesh
The close coupling relation that record brings.Specifically in implementation process, when a file is created, it is true that it reads its parent directory first
Timing is waited and can creatied directory, and is illustrated in method above, and the metadata of its parent directory can be by the catalogue of client
Caching is obtained.Be assigned to file on a file metadata node and stored by client, and this distribution method is based on one
The distribution method of cause property Hash.While storage, file is added into its institute in the directory content caching of this document metadata
Cached in the directory content of node, interacted with other nodes so as to avoid, reduce the delay of access.When need obtain
During directory content, it is only necessary to access each node, it is possible to which uncoupled directory content is aggregated into a complete catalogue
Appearance is sent to application program.
As shown in Fig. 2 in the implementation procedure of example, in addition it is also necessary to carry out persistence to metadata by key value database,
This example has used a kind of method that key value database stores metadata, and its method is with file or the title or the overall situation of catalogue
Unique mark, using the metadata of file or catalogue as value, is stored in key value database as key.
In directory metadata node, its key value database using the path of file as search for key, after
Metadata is stored in directory metadata node as value.
In file metadata node, the character string constituted with the GUID of parent directory and filename be for
The key of search, using other metadata as value, stores in file metadata node.
In the client of distributed file system, first number is cached using with directory metadata node identical key assignments mode
According to.Its specific process is, when the metadata of one catalogue of storage, using the path of catalogue as key, with other yuan of catalogue
Data are stored for value, meanwhile, add the GUID for catalogue, this identifier to have the unit of catalogue in the rear end of key assignments
Back end end is managed.In establishment file, using the unique mark and filename of its parent directory as key, its with file
Remaining metadata is stored as value.Now because file is distributed across each node, if being text with a unique node
Part is created and one unique identifier of management will inevitably make this node as performance bottleneck.Therefore, this example is used
The mark of the node of document creation and the metadata node of establishment file are the unique mark on this node of this document creation
Constitute a unique file identification for the overall situation.This mark file after is modified, such as renaming or shifting
Dynamic path will not all change.When the metadata of file is preserved, this mark can be stored the afterbody in key assignments.
In the internal data structure of key assignments storage, this example is deposited for different data, services using different key assignments
Storage.Wherein, using the key assignments data storage storehouse based on B-tree, file metadata node is used based on Hash directory metadata node
Key assignments data storage storehouse, in the client of distributed file system using based on internal memory database purchase catalogue metadata
Caching.
In the optimization stored for metadata, this example uses the key of random length and the value of fixed length.This method requirement
What the field of each metadata was to determine.Wherein, each value of the metadata fields of directory metadata is fixed length.File unit number
It is also fixed length according to each value of the metadata fields of node.The field of one fixed length is directly stored in key assignments and deposits when storage
In the value of storage, do not serialized and unserializing.In addition, this method does not need extra internal storage data in storage
Structure carrys out cache metadata, directly by metadata cache in the caching of key value database.
Realized in result of the invention, all of metadata operation for file at most accesses 2 minor nodes, in catalogue
Under the caching situation of metadata, it is only necessary to access a minor node, its time delay is only once to access round RTT, due to using
Key assignments is stored, thus it is very low for the time delay of metadata acquisition, be can be ignored in the RTT time delays of Ethernet, so
This method can be effectively reduced the information exchange between each node when distributed file system accesses metadata, reduce
The delay of metadata access, meanwhile, the method by separating directory content, decoupling strong association between file and catalogue
Property, handling capacity very high can be reached, so as to improve treatment effeciency of the distributed file system for metadata.
In addition, the decoupling location mode of the metadata of distributed type file system of the embodiment of the present invention other constitute and
Effect is all for a person skilled in the art known, in order to reduce redundancy, is not repeated.
In the description of this specification, reference term " one embodiment ", " some embodiments ", " example ", " specifically show
The description of example " or " some examples " etc. means to combine specific features, structure, material or spy that the embodiment or example are described
Point is contained at least one embodiment of the invention or example.In this manual, to the schematic representation of above-mentioned term not
Necessarily refer to identical embodiment or example.And, the specific features of description, structure, material or feature can be any
One or more embodiments or example in combine in an appropriate manner.
Although an embodiment of the present invention has been shown and described, it will be understood by those skilled in the art that:Not
Can these embodiments be carried out with various changes, modification, replacement and modification in the case of departing from principle of the invention and objective, this
The scope of invention is by claim and its equivalent limits.
Claims (9)
1. the decoupling location mode of a kind of metadata of distributed type file system, it is characterised in that comprise the following steps:
S1:Metadata to distributed file system is separated, to obtain the metadata of catalogue, the metadata of directory entry and text
The metadata of part;
S2:The metadata of the catalogue is arranged on directory inode;
S3:Each directory entry is split according to the distribution situation of file, and it is associated in the node storage of file storage
Directory entry, and set up point to directory metadata reverse indexing.
2. the decoupling location mode of metadata of distributed type file system according to claim 1, it is characterised in that described
Directory operation is included where establishment, the deletion of catalogue, all metadata for reading catalogue, obtaining catalogue, the change catalogue of catalogue
User's group and change catalogue belonging to user.
3. the decoupling location mode of metadata of distributed type file system according to claim 1, it is characterised in that also wrap
Include:
The globally unique mark for determining file is provided;
The cryptographic Hash of the global described mark of the file accessed required for calculating;
The node that metadata is deposited is positioned according to the cryptographic Hash.
4. the decoupling location mode of metadata of distributed type file system according to claim 3, it is characterised in that described
It is designated the fullpath of file.
5. the decoupling location mode of metadata of distributed type file system according to claim 1, it is characterised in that also wrap
Include:
When establishment file or catalogue, create one in the node of establishment file or catalogue and include the file or described
All directory entries in the parent directory path of catalogue;
If all or part of of the directory entry is created in the node, remaining directory entry is created.
6. the decoupling location mode of metadata of distributed type file system according to claim 1, it is characterised in that also wrap
Include:
When a file is deleted, node where deleting the metadata and the file of node where the file is corresponding
Directory entry metadata points to the project of the file.
7. the decoupling location mode of metadata of distributed type file system according to claim 1, it is characterised in that also wrap
Include:
When be read out catalogue or deltree operation when, access all of metadata node, read catalogue or deletion to obtain
All directory entries under catalogue.
8. the decoupling location mode of metadata of distributed type file system according to claim 1, it is characterised in that also wrap
Include:
Client-cache is provided, wherein, it is true when the directory metadata of the client-cache is for client establishment file
Determine the authority whether with establishment file;
The client accesses the metadata of catalogue, to obtain access rights when the metadata of file is accessed;
When the client has access rights, the metadata of file is accessed.
9. the decoupling location mode of metadata of distributed type file system according to claim 8, it is characterised in that also wrap
Include:
In the cache invalidation of described directory metadata client, the change of the authority of the catalogue of directory metadata and right is carried out
In the deletion of catalogue.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201710016284.7A CN106874383B (en) | 2017-01-10 | 2017-01-10 | Decoupling distribution method of metadata of distributed file system |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201710016284.7A CN106874383B (en) | 2017-01-10 | 2017-01-10 | Decoupling distribution method of metadata of distributed file system |
Publications (2)
Publication Number | Publication Date |
---|---|
CN106874383A true CN106874383A (en) | 2017-06-20 |
CN106874383B CN106874383B (en) | 2019-12-20 |
Family
ID=59165495
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201710016284.7A Active CN106874383B (en) | 2017-01-10 | 2017-01-10 | Decoupling distribution method of metadata of distributed file system |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN106874383B (en) |
Cited By (22)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN107479827A (en) * | 2017-07-24 | 2017-12-15 | 上海德拓信息技术股份有限公司 | A kind of mixing storage system implementation method based on IO and separated from meta-data |
CN108595287A (en) * | 2018-04-27 | 2018-09-28 | 新华三技术有限公司成都分公司 | Data truncation method and device based on correcting and eleting codes |
CN108984617A (en) * | 2018-06-13 | 2018-12-11 | 西安交通大学 | A kind of metadata catalog structure implementation method towards memory cloud |
CN109783449A (en) * | 2018-12-13 | 2019-05-21 | 深圳壹账通智能科技有限公司 | Data query processing method, platform, system and readable storage medium storing program for executing |
CN109783462A (en) * | 2018-12-13 | 2019-05-21 | 创新科存储技术有限公司 | A kind of data access method and device based on distributed file system |
CN109947730A (en) * | 2017-07-25 | 2019-06-28 | 中兴通讯股份有限公司 | Metadata restoration methods, device, distributed file system and readable storage medium storing program for executing |
CN110046133A (en) * | 2019-04-12 | 2019-07-23 | 苏州浪潮智能科技有限公司 | A kind of metadata management method, the apparatus and system of storage file system |
CN110704375A (en) * | 2019-09-26 | 2020-01-17 | 深圳前海大数金融服务有限公司 | File management method, device, equipment and computer storage medium |
CN110727675A (en) * | 2018-07-17 | 2020-01-24 | 阿里巴巴集团控股有限公司 | Method and device for processing linked list |
CN111143293A (en) * | 2019-12-22 | 2020-05-12 | 浪潮电子信息产业股份有限公司 | Metadata acquisition method, device, equipment and computer readable storage medium |
CN111258508A (en) * | 2020-02-16 | 2020-06-09 | 西安奥卡云数据科技有限公司 | Metadata management method in distributed object storage |
CN111309677A (en) * | 2020-02-11 | 2020-06-19 | 西安奥卡云数据科技有限公司 | File management method and device of distributed file system |
CN111400249A (en) * | 2020-03-06 | 2020-07-10 | 深圳市瑞驰信息技术有限公司 | File storage system and method easy for counting file number |
CN111949619A (en) * | 2020-07-21 | 2020-11-17 | 苏州元核云技术有限公司 | Dynamic directory generation method, system, electronic device and storage medium |
WO2021004295A1 (en) * | 2019-07-05 | 2021-01-14 | 中兴通讯股份有限公司 | Metadata processing method and apparatus, and computer-readable storage medium |
WO2021103600A1 (en) * | 2019-11-29 | 2021-06-03 | 浪潮电子信息产业股份有限公司 | Method, apparatus and device for deleting distributed system file, and storage medium |
CN113190505A (en) * | 2021-04-15 | 2021-07-30 | 网宿科技股份有限公司 | Metadata management method, file storage system and server |
CN113836143A (en) * | 2021-09-28 | 2021-12-24 | 新华三大数据技术有限公司 | Index creation method and device |
CN114048185A (en) * | 2021-11-18 | 2022-02-15 | 北京聚存科技有限公司 | Method for transparently packaging, storing and accessing massive small files in distributed file system |
CN115098466A (en) * | 2022-07-18 | 2022-09-23 | 重庆紫光华山智安科技有限公司 | Metadata management method and device, storage node and readable storage medium |
CN116795296A (en) * | 2023-08-16 | 2023-09-22 | 中移(苏州)软件技术有限公司 | Data storage method, storage device and computer readable storage medium |
US12001397B2 (en) | 2019-11-29 | 2024-06-04 | Inspur Electronic Information Industry Co., Ltd. | Method, apparatus and device for deleting distributed system file, and storage medium |
Citations (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN101354726A (en) * | 2008-09-17 | 2009-01-28 | 中国科学院计算技术研究所 | Method for managing memory metadata of cluster file system |
CN101692239A (en) * | 2009-10-19 | 2010-04-07 | 浙江大学 | Method for distributing metadata of distributed type file system |
CN103150394A (en) * | 2013-03-25 | 2013-06-12 | 中国人民解放军国防科学技术大学 | Distributed file system metadata management method facing to high-performance calculation |
CN103473337A (en) * | 2013-09-22 | 2013-12-25 | 北京航空航天大学 | Massive catalogs and files oriented processing method in distributed type storage system |
CN104537023A (en) * | 2014-12-19 | 2015-04-22 | 华为技术有限公司 | Storage method and device for reverse index records |
-
2017
- 2017-01-10 CN CN201710016284.7A patent/CN106874383B/en active Active
Patent Citations (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN101354726A (en) * | 2008-09-17 | 2009-01-28 | 中国科学院计算技术研究所 | Method for managing memory metadata of cluster file system |
CN101692239A (en) * | 2009-10-19 | 2010-04-07 | 浙江大学 | Method for distributing metadata of distributed type file system |
CN103150394A (en) * | 2013-03-25 | 2013-06-12 | 中国人民解放军国防科学技术大学 | Distributed file system metadata management method facing to high-performance calculation |
CN103473337A (en) * | 2013-09-22 | 2013-12-25 | 北京航空航天大学 | Massive catalogs and files oriented processing method in distributed type storage system |
CN104537023A (en) * | 2014-12-19 | 2015-04-22 | 华为技术有限公司 | Storage method and device for reverse index records |
Non-Patent Citations (2)
Title |
---|
PENG SUN: "MetaFlow: A Scalable Metadata Lookup Service for Distributed File Systems in Data Centers", 《IEEE TRANSACTIONS ON BIG DATA》 * |
YAO SUN: "A Distributed Cache Framework for Metadata Service of Distributed File Systems", 《2013 19TH IEEE INTERNATIONAL CONFERENCE ON PARALLEL AND DISTRIBUTED SYSTEMS》 * |
Cited By (31)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN107479827A (en) * | 2017-07-24 | 2017-12-15 | 上海德拓信息技术股份有限公司 | A kind of mixing storage system implementation method based on IO and separated from meta-data |
CN109947730A (en) * | 2017-07-25 | 2019-06-28 | 中兴通讯股份有限公司 | Metadata restoration methods, device, distributed file system and readable storage medium storing program for executing |
CN109947730B (en) * | 2017-07-25 | 2024-02-02 | 中兴通讯股份有限公司 | Metadata recovery method, device, distributed file system and readable storage medium |
CN108595287A (en) * | 2018-04-27 | 2018-09-28 | 新华三技术有限公司成都分公司 | Data truncation method and device based on correcting and eleting codes |
CN108595287B (en) * | 2018-04-27 | 2021-11-05 | 新华三技术有限公司成都分公司 | Data truncation method and device based on erasure codes |
CN108984617A (en) * | 2018-06-13 | 2018-12-11 | 西安交通大学 | A kind of metadata catalog structure implementation method towards memory cloud |
CN110727675A (en) * | 2018-07-17 | 2020-01-24 | 阿里巴巴集团控股有限公司 | Method and device for processing linked list |
CN110727675B (en) * | 2018-07-17 | 2023-06-27 | 阿里巴巴集团控股有限公司 | Method and device for processing linked list |
CN109783449A (en) * | 2018-12-13 | 2019-05-21 | 深圳壹账通智能科技有限公司 | Data query processing method, platform, system and readable storage medium storing program for executing |
CN109783462A (en) * | 2018-12-13 | 2019-05-21 | 创新科存储技术有限公司 | A kind of data access method and device based on distributed file system |
CN109783462B (en) * | 2018-12-13 | 2021-01-05 | 创新科技术有限公司 | Data access method and device based on distributed file system |
CN110046133A (en) * | 2019-04-12 | 2019-07-23 | 苏州浪潮智能科技有限公司 | A kind of metadata management method, the apparatus and system of storage file system |
WO2021004295A1 (en) * | 2019-07-05 | 2021-01-14 | 中兴通讯股份有限公司 | Metadata processing method and apparatus, and computer-readable storage medium |
CN110704375A (en) * | 2019-09-26 | 2020-01-17 | 深圳前海大数金融服务有限公司 | File management method, device, equipment and computer storage medium |
WO2021103600A1 (en) * | 2019-11-29 | 2021-06-03 | 浪潮电子信息产业股份有限公司 | Method, apparatus and device for deleting distributed system file, and storage medium |
US12001397B2 (en) | 2019-11-29 | 2024-06-04 | Inspur Electronic Information Industry Co., Ltd. | Method, apparatus and device for deleting distributed system file, and storage medium |
CN111143293A (en) * | 2019-12-22 | 2020-05-12 | 浪潮电子信息产业股份有限公司 | Metadata acquisition method, device, equipment and computer readable storage medium |
CN111143293B (en) * | 2019-12-22 | 2022-06-07 | 浪潮电子信息产业股份有限公司 | Metadata acquisition method, device, equipment and computer readable storage medium |
CN111309677A (en) * | 2020-02-11 | 2020-06-19 | 西安奥卡云数据科技有限公司 | File management method and device of distributed file system |
CN111309677B (en) * | 2020-02-11 | 2023-05-23 | 西安奥卡云数据科技有限公司 | File management method and device of distributed file system |
CN111258508A (en) * | 2020-02-16 | 2020-06-09 | 西安奥卡云数据科技有限公司 | Metadata management method in distributed object storage |
CN111400249A (en) * | 2020-03-06 | 2020-07-10 | 深圳市瑞驰信息技术有限公司 | File storage system and method easy for counting file number |
CN111949619B (en) * | 2020-07-21 | 2024-04-26 | 苏州元核云技术有限公司 | Dynamic catalog generation method, system, electronic equipment and storage medium |
CN111949619A (en) * | 2020-07-21 | 2020-11-17 | 苏州元核云技术有限公司 | Dynamic directory generation method, system, electronic device and storage medium |
CN113190505A (en) * | 2021-04-15 | 2021-07-30 | 网宿科技股份有限公司 | Metadata management method, file storage system and server |
CN113836143A (en) * | 2021-09-28 | 2021-12-24 | 新华三大数据技术有限公司 | Index creation method and device |
CN113836143B (en) * | 2021-09-28 | 2024-02-27 | 新华三大数据技术有限公司 | Index creation method and device |
CN114048185A (en) * | 2021-11-18 | 2022-02-15 | 北京聚存科技有限公司 | Method for transparently packaging, storing and accessing massive small files in distributed file system |
CN115098466A (en) * | 2022-07-18 | 2022-09-23 | 重庆紫光华山智安科技有限公司 | Metadata management method and device, storage node and readable storage medium |
CN116795296B (en) * | 2023-08-16 | 2023-11-21 | 中移(苏州)软件技术有限公司 | Data storage method, storage device and computer readable storage medium |
CN116795296A (en) * | 2023-08-16 | 2023-09-22 | 中移(苏州)软件技术有限公司 | Data storage method, storage device and computer readable storage medium |
Also Published As
Publication number | Publication date |
---|---|
CN106874383B (en) | 2019-12-20 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN106874383A (en) | A kind of decoupling location mode of metadata of distributed type file system | |
US8214334B2 (en) | Systems and methods for distributed system scanning | |
US8296312B1 (en) | Search and update of attributes in file systems | |
CA2139693C (en) | Summary catalogs | |
US9043372B2 (en) | Metadata subsystem for a distributed object store in a network storage system | |
US7430570B1 (en) | Shadow directory structure in a distributed segmented file system | |
US7228299B1 (en) | System and method for performing file lookups based on tags | |
US8214400B2 (en) | Systems and methods for maintaining distributed data | |
US8176013B2 (en) | Systems and methods for accessing and updating distributed data | |
CN103282899B (en) | The storage method of data, access method and device in file system | |
US20130013602A1 (en) | Database system | |
CN113010486B (en) | Metadata layered caching method and device for centerless distributed file system | |
US20040225963A1 (en) | Dynamic maintenance of web indices using landmarks | |
US9081784B2 (en) | Delta indexing method for hierarchy file storage | |
CN105824723A (en) | Method and system for backup of data of public cloud storage account | |
US7844596B2 (en) | System and method for aiding file searching and file serving by indexing historical filenames and locations | |
WO2023179787A1 (en) | Metadata management method and apparatus for distributed file system | |
CN103136294B (en) | File operating method and device | |
JP3300399B2 (en) | Computer system and file management method | |
CN103246718B (en) | File access method, device and equipment | |
CN105468599A (en) | Metadata hierarchy management method for storage virtualization system | |
CN110019016A (en) | The KV for providing logic key stores device and method thereof | |
WO2021004295A1 (en) | Metadata processing method and apparatus, and computer-readable storage medium | |
JPH08235040A (en) | Data file management system | |
CN117435559B (en) | Metadata hierarchical management method and device, storage medium and electronic equipment |
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 | ||
CB03 | Change of inventor or designer information |
Inventor after: Lu Youyou Inventor after: Shu Jiwu Inventor before: Lu Youyou Inventor before: Shu Jiwu Inventor before: Li Siyang |
|
CB03 | Change of inventor or designer information |