WO2014134905A1 - 文件系统的属性同步控制方法、装置和系统 - Google Patents

文件系统的属性同步控制方法、装置和系统 Download PDF

Info

Publication number
WO2014134905A1
WO2014134905A1 PCT/CN2013/082289 CN2013082289W WO2014134905A1 WO 2014134905 A1 WO2014134905 A1 WO 2014134905A1 CN 2013082289 W CN2013082289 W CN 2013082289W WO 2014134905 A1 WO2014134905 A1 WO 2014134905A1
Authority
WO
WIPO (PCT)
Prior art keywords
attribute
board
update
synchronization
metadata
Prior art date
Application number
PCT/CN2013/082289
Other languages
English (en)
French (fr)
Inventor
朱鹏
林健
俞超
胡剑华
Original Assignee
中兴通讯股份有限公司
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by 中兴通讯股份有限公司 filed Critical 中兴通讯股份有限公司
Publication of WO2014134905A1 publication Critical patent/WO2014134905A1/zh

Links

Classifications

    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06FELECTRIC DIGITAL DATA PROCESSING
    • G06F16/00Information retrieval; Database structures therefor; File system structures therefor
    • G06F16/10File systems; File servers
    • G06F16/17Details of further file system functions
    • G06F16/178Techniques for file synchronisation in file systems

Definitions

  • the present invention relates to the field of communications, and in particular to a method, device and system for attribute synchronization control of a file system.
  • BACKGROUND With the rapid development of the multimedia industry, due to various considerations such as cost and reliability, more and more manufacturers choose to deploy distributed upper-layer storage systems in products, and the distributed file system has also been rapidly developed. . At present, the common network file system pays attention to the efficiency of data caching, and ignores the speed and performance improvement of the attribute cache which is operated more frequently.
  • the attributes in the distributed file system are stored on a reliable node, and other nodes. When the attribute is operated, the reliable node needs to be accessed.
  • the present invention provides a method, an apparatus, and a system for controlling attribute synchronization of a file system to solve the above problems.
  • a method for controlling attribute synchronization of a file system including: the application board determines a synchronization mode according to a current type of update operation on an attribute; wherein the update operation type includes: simple operation and complexity Operation: The application board performs attribute synchronization operations on the metadata board of the attribute according to the determined synchronization mode, and stores the synchronized attributes locally.
  • the method for determining the synchronization according to the current update operation type of the attribute includes: If the current update operation type of the attribute is simple operation, the application board determines that the synchronization mode is an update unrestricted synchronization operation; if the current attribute is updated
  • the operation type is a complex operation.
  • the application board determines that the synchronization mode is the update-limited synchronization operation.
  • the application board performs the attribute synchronization operation on the metadata board of the attribute according to the determined synchronization mode.
  • the method is as follows: When the synchronization mode determined by the application board is an update unrestricted synchronization operation, the application board updates the local attributes. And sending the pre-updated attribute and the updated attribute to the attribute's metadata board; the application board receives the response message of the metadata board, and if the response message indicates that the update operation is successful, the application board uses the updated attribute as The new attribute; if the response message indicates that the update operation fails, the application board takes the attribute carried in the response message as a new attribute.
  • the application board performs the attribute synchronization operation on the metadata board of the attribute according to the determined synchronization mode.
  • the requesting board sends the attribute to the metadata board of the attribute when the synchronization mode determined by the application board is the update-limited synchronization operation.
  • the requesting board obtains the request for the update permission; the application board receives the indication of the update authority and the corresponding attribute returned by the metadata board; the application board updates the attribute, and periodically sends the updated attribute to the metadata board.
  • the method further includes: after the requesting board receives the request for acquiring the attribute of the third board, and sends the updated attribute to the third board.
  • the method further includes: the application board records the address and the acquisition time of the third board; the application board determines the time of acquiring the third board according to the record. Whether the number of acquisitions of the third board in the specified duration is greater than the threshold. If yes, the third board is marked as a hot node of the attribute. After the attribute is updated, the updated attribute is sent to the hot node of the attribute. .
  • a method for controlling attribute synchronization of a file system including: a metadata board receiving an update request of an application board; and a metadata board determining a synchronization manner according to information carried in the update request.
  • the metadata board performs attribute synchronization with the application board according to the determined synchronization mode, and stores the synchronized attributes locally.
  • the data board determines the synchronization manner according to the information carried in the update request, and includes: if the information carried in the update request is an attribute before the update of the application board and the updated attribute, the metadata board determines that the synchronization mode is an update.
  • the metadata board determines that the synchronization mode is a synchronization-limited synchronization operation.
  • the metadata synchronization board performs the attribute synchronization operation with the application board according to the determined synchronization mode.
  • the metadata board compares the application board before the update of the application board. If the attribute is the same as the local one, the metadata board sends a response message with the update operation success indication to the application board.
  • the updated attribute of the application board is used as the new attribute and is added to the application board. If the board is inconsistent, the metadata board sends a response message carrying the update operation failure indication and the local attribute to the application board.
  • the metadata synchronization board performs the attribute synchronization with the application board according to the determined synchronization mode.
  • the metadata board determines whether to grant the application board update permission when the synchronization mode is the update-limited synchronization operation. If the certificate is sent, the metadata board sends the update permission permission indication and the corresponding attribute to the application board, and broadcasts the address of the application board and sets the cache attribute to invalid notification to the board except the application board.
  • the metadata board periodically receives the updated attributes of the application board and uses the updated attributes as new attributes.
  • the method further includes: when one of the following occurs, the metadata board cancels the update permission of the application board: the metadata board receives the attribute update completion notification sent by the application board; the metadata board is within the set duration The message from the application board is not received.
  • an attribute synchronization control apparatus for a file system including: a first synchronization mode determining module, configured to determine a synchronization mode according to a current update operation type of an attribute; wherein the update operation The type includes: simple operation and complex operation; the first synchronous operation module is configured to perform the attribute synchronization operation on the metadata board of the attribute determined by the synchronization mode determined by the first synchronization mode, and store the synchronized attribute locally.
  • the first synchronization mode determining module includes: a first synchronization mode determining unit, configured to: if the current update operation type of the attribute is a simple operation, determining that the synchronization mode is an update unrestricted synchronization operation; the second synchronization mode determining unit, setting If the current update operation type for the attribute is a complex operation, it is determined that the synchronization mode is an update-limited synchronization operation.
  • the first synchronization operation module includes: an update attribute sending unit, configured to: when the synchronization mode determined by the application board is an update unrestricted synchronization operation, update the local attribute, and update the attribute and the update status The attribute is sent to the metadata board of the attribute; the first new attribute determining unit is configured to receive the response message of the metadata board, and if the response message indicates that the update operation is successful, the updated attribute is used as a new attribute; if the response message indicates The update operation failed, and the attribute carried in the response message is taken as a new attribute.
  • the first synchronization operation module includes: a permission requesting unit, configured to send a request for obtaining an update permission to the attribute metadata board when the synchronization mode determined by the application board is a synchronization operation with limited restriction; the information receiving unit, setting And receiving an indication of the update authority and an attribute corresponding to the indication returned by the metadata board; the second update attribute sending unit is configured to perform an update operation on the attribute received by the information receiving unit, and periodically send the updated attribute to the metadata veneer.
  • the module is configured to perform the attribute synchronization operation with the application board according to the synchronization mode determined by the second synchronization mode, and store the synchronized attributes locally.
  • the second synchronization mode determining module includes: a third synchronization mode determining unit, configured to: if the information carried in the update request is an attribute before the update of the application board and the updated attribute, determining that the synchronization mode is an update operation that is not limited by the update And a fourth synchronization mode determining unit, configured to determine that the synchronization mode is a synchronization-limited synchronization operation if the information carried in the update request is a request to obtain an update authority.
  • the second synchronization operation module includes: an attribute comparison unit, configured to compare whether the attribute before the update of the application board is consistent with the local attribute when the synchronization mode determined by the first synchronization mode determining unit is an update operation that is not restricted;
  • the processing unit is configured to send a response message carrying the update operation success indication to the application board as the new attribute, and send the attribute of the application board as a new attribute, and send the attribute to the application board. If the comparison result of the attribute comparison unit is inconsistent, the response message carrying the update operation failure indication and the local attribute is sent to the application board.
  • the second synchronization operation module includes: a permission determination unit, configured to determine whether to grant an application board update permission when the synchronization mode determined by the second synchronization mode determination unit is a synchronization operation with limited update;
  • the judgment result of the judgment unit is the grant, and the indication corresponding to the update authority and the attribute corresponding to the indication are sent to the application board, and the address of the application board and the notification that the cache attribute is invalid are not broadcast to the board except the application board.
  • the attribute update unit is set to periodically receive the updated attribute of the application board, and use the updated attribute as a new attribute.
  • an attribute synchronization control system of a file system including an application board and a metadata board, wherein the application board includes the attribute synchronization in another aspect of the foregoing embodiment of the present invention.
  • the control device, the metadata veneer includes the attribute synchronization control device in a further aspect of the above embodiment of the present invention.
  • the attribute update operation type is used to perform the attribute synchronization process with the metadata board, thereby simplifying the attribute synchronization process, and storing the synchronized attributes on the application board side for subsequent
  • the application board provides a convenient way to obtain this attribute. This method avoids the problem that the attribute is stored in a reliable node and causes system performance degradation.
  • FIG. 1 is a flowchart of an attribute synchronization control method of a file system described from the side of an application board according to an embodiment of the present invention
  • FIG. 2 is a state transition state diagram of a single attribute in an attribute cache according to an embodiment of the present invention
  • FIG. 4 is a flowchart of a method for synchronizing a simple modified attribute according to an embodiment of the present invention
  • FIG. 5 is a flowchart according to an embodiment of the present invention
  • FIG. 6 is a structural block diagram of an attribute synchronization control apparatus of a file system described from the side of an application board according to an embodiment of the present invention
  • FIG. 6 is a slave metadata according to an embodiment of the present invention.
  • FIG. 8 is a block diagram showing the structure of the attribute synchronization control system of the file system according to an embodiment of the present invention.
  • a file system in the embodiment of the present invention is configured with a plurality of boards (also referred to as nodes).
  • the board that stores the metadata is called a metadata board, and the board that performs the attribute operation is called an application.
  • the attribute synchronization control method of the file system is described from the side of the application board and the metadata board.
  • a flowchart of the attribute synchronization control method of the file system shown in FIG. 1 is described by taking the description from the application board side as an example. The method includes the following steps: Step S102: The application board updates the current attribute according to the current operation.
  • the operation is classified into a complex operation; or the current operation needs to modify the attributes of the plurality of metadata, and the operation may be classified into a complex operation.
  • the application board performs the synchronization mode and the attribute metadata board determined above. The property synchronizes operations and stores the synchronized properties locally.
  • the attribute update operation type is used to perform the attribute synchronization process with the metadata board, thereby simplifying the attribute synchronization process, and storing the synchronized attributes on the application board side for subsequent application boards. Obtaining this attribute provides a convenient way to avoid the problem that the attribute is stored in a reliable node and the performance of the system is degraded. The performance and resource consumption caused by the access and modification of the attribute are reduced, and the system performance is optimized. If the update operation type of the current application card is simple, the application board does not need to obtain the update permission from the metadata board. The application board can determine that the synchronization mode is an update-unlimited synchronization operation.
  • the update operation type is a complex operation.
  • the application board in this embodiment can determine that the synchronization mode is a synchronization-limited synchronization operation.
  • the following describes the update unrestricted synchronization operation and the update limited synchronization operation, respectively.
  • I. Update unrestricted synchronization operation When the synchronization mode determined by the application board is to update the unrestricted synchronization operation, the application board first updates the local attributes, and the attributes before the update (also called The original attribute and the updated attribute are sent to the metadata board of the attribute. The application board receives the response message of the metadata board. If the response message indicates that the update operation is successful, the application board updates the updated attribute as a new one.
  • the application board takes the attribute carried in the response message as a new attribute. In this mode, the application board will save the new attribute locally. When the application board needs to access the attribute next time, the attribute can be obtained locally. To ensure the validity of the attribute, you can This attribute sets a status flag. If the status flag is I, the attribute is an invalid attribute. Otherwise, the attribute is valid. 2. Update the unrestricted synchronization operation. When the synchronization mode determined by the application board is the update-restricted synchronization operation, the application board sends a request for obtaining the update permission to the metadata board of the attribute.
  • the application board receives the indication of the update permission returned by the metadata board and the attribute corresponding to the indication; the application board updates the attribute, and periodically sends the updated attribute (that is, the new attribute) to the metadata board. .
  • the application board will store the updated attributes locally. Because this complex operation mode performs multiple operations on the attributes, this embodiment uses periodic updates to the metadata boards. The way to make the attributes on the metadata board the latest attributes.
  • the embodiment of the present invention considers that there are many boards in the system. In order to facilitate the proximity and decentralized access attributes of other boards, this embodiment uses the MOSIE (Modifier Owner Share Exclusive Invalid) method of modifying, owning, sharing, exclusive, and invalid.
  • MOSIE Modifier Owner Share Exclusive Invalid
  • 0 Refers to the owner of the attribute, only has read-only permission (currently where the metadata is located, and records the location of M); S: the sharer of the attribute, the node only has the read-only permission of the attribute;
  • the attribute on the node of this state is invalid and meaningless.
  • the metadata board will notify the other nodes of the address of the application board.
  • the third board can save the address of the application board.
  • the board can send a request.
  • the foregoing method further includes: after the requesting board receives the request for acquiring the attribute by the third board, and sending the updated attribute to the third board.
  • the third board can conveniently obtain the foregoing attributes, and can avoid a large number of nodes frequently accessing the metadata board, thereby reducing the pressure on the metadata board.
  • the application board sends the updated attributes to the third board.
  • the method further includes: requesting the board to record the address and acquisition time of the third board;
  • the acquisition time of the third board determines whether the number of times the third board is acquired within a specified duration is greater than a threshold. If yes, the third board is marked as a hot node of the attribute, and then used to perform the update operation on the attribute. Will be updated
  • the properties are sent directly to the hot node of the property. In this way, the hot node of the attribute does not need to send the attribute to the requesting node, so that the hot node of the attribute can obtain the updated attribute in time.
  • a flowchart of the attribute synchronization control method of the file system shown in FIG. 3 is described by taking the description from the side of the metadata board as an example.
  • Step S302 The metadata board receives the application board. And updating the request; step S304, the metadata board determines the synchronization mode according to the information carried in the update request; Step S306, the metadata board performs the attribute synchronization operation with the application board according to the determined synchronization manner, and after the local storage synchronization Attributes.
  • the synchronization mode is determined according to the information in the update request of the application board, and the attribute synchronization process is performed on the application board in a targeted manner, which simplifies the attribute synchronization process, and the synchronized attribute is on the metadata board.
  • the storage on the side provides a convenient way for other boards to obtain the attribute.
  • the metadata board may determine that the synchronization mode is an update operation that is not restricted by the update; if the update request is carried in the update request The information is a request for obtaining the update permission, and the metadata board can determine that the synchronization mode is a synchronization operation with limited update. If the synchronization mode of the metadata board is the same as that of the update, the metadata board compares the attributes of the preceding application board with the local attributes.
  • the metadata board sends the data to the application board.
  • the response message carrying the update operation success indication is used as the new attribute, and the new attribute is broadcast to the board except the application board.
  • the metadata board is applied to the application board.
  • a response message carrying an update operation failure indication and a local attribute is sent. This makes it easy to update the properties.
  • the synchronization mode determined by the metadata board is the update-limited synchronization operation
  • the metadata board determines whether the application board is granted the update permission.
  • the specific judgment mode can be combined with the specific content of the current operation to determine whether the application node is suitable for the operation.
  • the metadata board sends the update permission permission indication and the corresponding attribute to the application board, and broadcasts the address of the application board and sets the cache attribute to the board except the application board. Invalid notification;
  • the metadata board can periodically receive the updated attributes of the application board and use the updated attributes as new attributes.
  • the method further includes: when one of the following occurs, the metadata board cancels the update permission of the application board: 1) the metadata board receives the attribute update completion notification sent by the application board; 2) the metadata list Board in The message from the application board is not received within the set time period; 3) the update permission request of the metadata board to the third board is confirmed, and the attribute update of the third board is better than the attribute of the application board. Update.
  • the update permission of the application board is cancelled in time, and the other boards can obtain the update authority, which facilitates the synchronous operation of the attributes.
  • the update permission of the application board is cancelled in time, and the other boards can obtain the update authority, which facilitates the synchronous operation of the attributes.
  • only the update of the attribute is described. In the system corresponding to the foregoing method, the following operations may be performed on the attribute:
  • Update attribute Generally, it refers to the modification attribute. As described in the above embodiment, the embodiment of the present invention involves two forms: a simple mode and a complex mode, that is, a synchronous operation and an update-limited synchronization corresponding to the above-mentioned update. operating. 3) Release the update permissions: Return the M permissions, then make an update to the universe.
  • Step S402 Simply modify a board to simultaneously send the original attribute and the modified attribute (ie, the new attribute) to the owner 0 of the attribute (ie, the element) Data board);
  • Step S404 the metadata board (referred to as "0 board” or "0 node") is compared with the original attribute sent by the received simple modified board according to the attribute of the record itself, if the two are consistent And saving the new attribute, and performing step S406; if not, performing step S410; step S406, the metadata board sends a modification success notification to the simple modification board, and then performing step S408; step S408, the metadata board is in the broadcast domain
  • the board broadcasts the above new attribute; wherein, the order of step S406 and step S408 described above may be interchanged;
  • Step S502 The complex modification board obtains the modification authority M from the metadata board, and carries the original attribute when sending the M request.
  • step S504 the metadata board compares with the original attribute sent by the received complex modification board according to the attribute recorded by itself, and if the same, and the complex modification board is suitable for granting M, step S506 is performed; otherwise, step S510 is performed; S506, the metadata board is assigned a complex modification board M, records the address of the complex modification board, and returns all the data of the own attribute information to the complex modification board, and then performs step S508; step S508, the metadata list
  • the board sends a notification of the update M to the board in the broadcast domain, and the notification carries the address of the complex modification board and the indication that the cache attribute is invalid (Invalid); that is, the update M node information is sent to the non-zero broadcast node, and the notification is invalid.
  • the sequence of the foregoing step S506 and the step S508 may be interchanged.
  • the complex modification board obtains the modification authority M, which may be referred to as an M node or an M board.
  • the complex modification board will be responsible for updating the relevant modifications as the update party of the attribute. Specifically include:
  • this method is also the heartbeat communication mode for maintaining the M state. If the heartbeat is continuously lost (that is, the heartbeat communication is received within the set time of the metadata board), the metadata is received. The board will complete the M state of the board and modify the M state of the board in the broadcast domain of the board. The related information is updated. At the same time, the metadata board will release the M state of the local record (that is, the board in the cluster loses the M board, and all updates and reads will re-initiate the request to the data 0 (that is, the metadata board)) In this embodiment, a protection period of 30 s is set. If the board is in the Ms for 30 seconds, the board cannot be applied to the M state. If the board is modified, the board can be re-applied. Permissions, the system will update the relevant information);
  • the complex modification board does not have any abnormality during the update. After the complex modification board completes the corresponding operation, the M state needs to be removed. For example: First, the attributes of the boards in all broadcast domains are updated, and the M status is sent to the metadata board. The notification that the board that receives the update attribute broadcast can send a confirmation update message to the metadata board. At this time, the metadata board will serve as the key node for synchronization, and ensure that the message in the broadcast domain used is received. In the step S510, the metadata board returns an application failure notification to the complex modification board, where the notification carries the attribute recorded on the metadata board and the address of the board that has the M authority (of course, if there is no board in the system with M Permission, there is no such address item in the notification).
  • the above process is a complex attribute modification process, which aims to improve the local response speed during continuous operations such as file system creation and writing.
  • the above-mentioned simple modification board or complex modification board can use broadcast, multicast, or single-wave mode when performing attribute synchronization with other nodes. For example, if the number of synchronization nodes needs to exceed the specified number, the broadcast mode can be adopted.
  • the attribute state of the node that needs to be synchronized is 8, if there is no node with M authority, then it is synchronized by 0 node), usually the first and second layer nodes in the system (assuming the current file system is mounted under /mnt, then /mnt/a This file or directory is the first layer node, and the /mnt/a/b file directory is the second layer node.
  • the cache design is tree-shaped, for each cache tree, the root of the tree is basically Each board exists, and the two layers of nodes can be specially processed during design. The default is broadcast.
  • the embodiment is preferably processed in the following manner:
  • the 0 node will initiate the action of the current M-state node to cancel the M-state to ensure the fairness of each node; (ie: the original 0-node requires the process of reclaiming the M-privilege, after this process all M requesting a board (complex request) will degenerate to a simple request, which is a general process without special processing).
  • the M is The board needs to be synchronized to the 0 node; at this time, the 0 node applies for the memory again, and re-synchronizes the modified attributes of the M board (the data on the disk does not need to be read at this time);
  • the embodiment of the present invention further provides a file system attribute synchronization control device, which may be disposed on the application board.
  • the device includes: a first synchronization mode determining module 62, setting The synchronization mode is determined according to the current update operation type of the attribute; wherein the update operation type includes: simple operation and complex operation; the first synchronization operation module 64 is connected to the first synchronization mode determination module 62, and is set to follow the first synchronization mode.
  • the synchronization mode determined by the determining module 62 performs an attribute synchronization operation with the attribute metadata board, and stores the synchronized attributes locally.
  • the attribute update operation type is used to perform the attribute synchronization process with the metadata board, thereby simplifying the attribute synchronization process, and storing the synchronized attributes on the application board side for subsequent application boards. Obtaining this attribute provides a convenient way to avoid the problem that the attribute is stored in a reliable node and the performance of the system is degraded. The performance and resource consumption caused by the access and modification of the attribute are reduced, and the system performance is optimized.
  • the first synchronization mode determining module 62 includes: a first synchronization mode determining unit, configured to: if the current update operation type of the attribute is a simple operation, determining that the synchronization mode is an update unrestricted synchronization operation;
  • the step mode determining unit is configured to determine that the synchronization mode is an update-limited synchronization operation if the current update operation type of the attribute is a complex operation.
  • the first synchronization operation module 64 includes: an update attribute sending unit, configured to: when the synchronization mode determined by the application board is an update unrestricted synchronization operation, update the local attribute, and update the attribute before the update And the updated attribute is sent to the metadata board of the attribute; the first new attribute determining unit is configured to receive the response message of the metadata board, and if the response message indicates that the update operation is successful, the updated attribute is used as a new attribute; The response message indicates that the update operation failed, and the attribute carried in the response message is taken as a new attribute.
  • the first synchronization operation module 64 includes: a permission requesting unit, configured to send a request for obtaining an update authority to the attribute metadata board when the synchronization mode determined by the application board is a synchronization operation with limited restriction; a unit, configured to receive an indication of the update permission returned by the metadata board and an attribute corresponding to the indication; the second update attribute sending unit is configured to perform an update operation on the attribute received by the information receiving unit, and periodically send the updated attribute Give the metadata board.
  • the embodiment of the present invention further provides a file system attribute synchronization control device, which may be disposed on the metadata board. Referring to FIG.
  • the device includes: an update request receiving module 72, configured to Receiving the update request of the application board; the second synchronization mode determining module 74 is connected to the update request receiving module 72, and is configured to determine the synchronization mode according to the information carried in the update request received by the update request receiving module 72.
  • the second synchronization operation module 76 And being connected to the second synchronization mode determining module 74, configured to perform the attribute synchronization operation with the application board according to the synchronization mode determined by the second synchronization mode determining module 74, and store the synchronized attributes locally.
  • the synchronization mode is determined according to the information in the update request of the application board, and the attribute synchronization process is performed on the application board in a targeted manner, which simplifies the attribute synchronization process, and the synchronized attribute is on the metadata board.
  • the storage on the side provides a convenient way for other boards to obtain the attribute. This method avoids the problem that the attribute is stored in a reliable node and the performance of the system is degraded. The performance and resources brought by the access and modification of the attribute are reduced. Consumption, optimized system performance.
  • the second synchronization mode determining module 74 includes: a third synchronization mode determining unit, configured to: if the information carried in the update request is an attribute before the application board is updated and an updated attribute, determining that the synchronization mode is an update is not limited
  • the synchronization operation unit is configured to: if the information carried in the update request is a request for obtaining the update authority, determine that the synchronization mode is a synchronization operation with limited update.
  • the second synchronization operation module 76 includes: an attribute comparison unit, configured to compare whether the attribute and the local attribute before the update of the application board are updated when the synchronization mode determined by the first synchronization mode determination unit is an update operation that is not restricted.
  • the synchronization processing unit is configured to send a response message carrying the update operation success indication to the application board as a new attribute, and to remove the application form, if the comparison result of the attribute comparison unit is the same.
  • the board other than the board broadcasts a new attribute. If the comparison result of the attribute comparison unit is inconsistent, the response message carrying the update operation failure indication and the local attribute is sent to the application board.
  • the second synchronization operation module 76 includes: a permission determination unit, configured to determine whether to grant an application board update permission when the synchronization mode determined by the second synchronization mode determination unit is a synchronization operation with limited update; If the judgment result of the authority judging unit is the grant, the application board sends an indication of the update authority and the attribute corresponding to the indication, and broadcasts the address of the application board and sets the cache attribute to the board except the application board. Invalid notification; The attribute update unit is set to periodically receive the updated attribute of the application board, and the updated attribute is used as the new attribute.
  • the embodiment of the present invention further provides an attribute synchronization control system for a file system, and FIG.
  • FIG. 8 is a structural block diagram of an attribute synchronization control system of a file system according to an embodiment of the present invention, as shown in FIG.
  • the application board 60 and the metadata board 70 are included, wherein the application board 60 includes the apparatus shown in FIG. 6 above, and the metadata board 70 includes the apparatus shown in FIG.
  • the system can significantly improve the performance of local access, and in the case of high-pressure access, the processing pressure of the system metadata board can be dispersed through policies.
  • the state in the above embodiment refers to a state corresponding to an attribute, and there may be hundreds of attributes on a node, and each attribute corresponds to a corresponding state.
  • the embodiment of the present invention manages the attributes of the entire file system by using an efficient synchronization manner, so that the remote client can determine whether the cached metadata is available according to the recorded data of the system. Therefore, the performance of the local node can be improved, and at the same time, the system can be relied upon to reduce the performance and resource consumption caused by the access and modification of the attribute, thereby avoiding the system performance caused by each node in the system acquiring the attribute from one node. Falling problems. Obviously, those skilled in the art should understand that the above modules or steps of the embodiments of the present invention can be implemented by a general computing device, which can be concentrated on a single computing device or distributed in multiple computing devices.
  • embodiments of the invention are not limited to any specific combination of hardware and software.
  • the above is only the preferred embodiment of the present invention, and is not intended to limit the present invention, and various modifications and changes can be made to the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and scope of the present invention are intended to be included within the scope of the present invention.
  • the technical solution provided by the embodiments of the present invention can be applied to the field of storage systems, and solves the problem that system performance is degraded when stored in a reliable node, and implements more frequent operations under the premise of ensuring efficient data caching. The speed and performance of the cache is improved.

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)
  • Information Transfer Between Computers (AREA)

Abstract

本发明公开了一种文件系统的属性同步控制方法、装置和系统。其中,该方法包括:申请单板根据当前对属性的更新操作类型确定同步方式(S102),其中,该更新操作类型包括:简易操作和复杂操作;申请单板按照确定的同步方式与属性的元数据单板进行属性同步操作,并在本地存储同步后的属性(S104)。通过本发明,避免了属性存储于一个可靠节点导致系统性能下降的问题,降低了属性的访问和修改所带来的性能和资源的消耗,优化了系统性能。

Description

文件系统的属性同步控制方法、 装置和系统 技术领域 本发明涉及通信领域, 具体而言, 涉及一种文件系统的属性同步控制方法、 装置 和系统。 背景技术 随着多媒体产业的迅猛发展, 出于成本、 可靠性等多方面的考虑, 越来越多的厂 商选择在产品中部署分布式上层存储系统,分布式文件系统也因此得到了快速的发展。 目前, 普通网络文件系统注重数据缓存的高效性, 而忽略了操作更为频繁的属性 缓存的速度与性能的提高, 通常, 分布式文件系统中的属性会存储在一个可靠的节点 上, 其它节点对该属性进行操作时, 都需要访问该可靠节点, 如果当前节点与该可靠 节点相距较远, 则访问所花时间较长; 并且如果系统内的各个节点均频繁访问该可靠 节点, 也会造成该可靠节点的响应压力较大。 由此可见, 将属性单纯地存储在一个可 靠节点上的方式影响了系统的性能。 针对相关技术中属性存储于一个可靠节点导致系统性能下降的问题, 目前尚未提 出有效的解决方案。 发明内容 针对上述属性存储于一个可靠节点导致系统性能下降的问题, 本发明实施例提供 了一种文件系统的属性同步控制方法、 装置和系统, 以至少解决上述问题。 根据本发明实施例的一个方面,提供了一种文件系统的属性同步控制方法,包括: 申请单板根据当前对属性的更新操作类型确定同步方式; 其中, 该更新操作类型包括: 简易操作和复杂操作; 申请单板按照确定的同步方式与属性的元数据单板进行属性同 步操作, 并在本地存储同步后的属性。 上述申请单板根据当前对属性的更新操作类型确定同步方式包括: 如果当前对属 性的更新操作类型为简易操作, 申请单板确定同步方式为更新不受限的同步操作; 如 果当前对属性的更新操作类型为复杂操作, 申请单板确定同步方式为更新受限的同步 操作。 上述申请单板按照确定的同步方式与属性的元数据单板进行属性同步操作包括: 当申请单板确定的同步方式为更新不受限的同步操作时, 申请单板对本地的属性进行 更新操作, 并将更新前的属性和更新后的属性发送给属性的元数据单板; 申请单板接 收元数据单板的响应消息, 如果响应消息指示更新操作成功, 申请单板将更新后的属 性作为新属性; 如果该响应消息指示更新操作失败, 申请单板将响应消息中携带的属 性作为新属性。 上述申请单板按照确定的同步方式与属性的元数据单板进行属性同步操作包括: 当申请单板确定的同步方式为更新受限的同步操作时, 申请单板向属性的元数据单板 发送获取更新权限的请求; 申请单板接收元数据单板返回的授予更新权限的指示和指 示对应的属性; 申请单板对属性进行更新操作, 并定期将更新后的属性发送给元数据 单板。 上述申请单板对属性进行更新操作之后, 上述方法还包括: 申请单板接收到第三 单板获取属性的请求后, 将更新后的属性发送给第三单板。 上述申请单板将更新后的属性发送给第三单板之前, 该方法还包括: 申请单板记 录第三单板的地址和获取时间; 申请单板根据记录的第三单板的获取时间判断第三单 板在指定时长内的获取次数是否大于阈值, 如果是, 标记第三单板为属性的热节点, 用以后续对属性进行更新操作之后, 将更新后的属性发送给属性的热节点。 根据本发明实施例的另一个方面, 提供了一种文件系统的属性同步控制方法, 包 括: 元数据单板接收申请单板的更新请求; 元数据单板根据更新请求中携带的信息确 定同步方式; 元数据单板按照确定的同步方式与申请单板进行属性同步操作, 并在本 地存储同步后的属性。 上述数据单板根据更新请求中携带的信息确定同步方式包括: 如果更新请求中携 带的信息是申请单板更新前的属性和更新后的属性, 元数据单板确定同步方式为更新 不受限的同步操作; 如果更新请求中携带的信息是获取更新权限的请求, 元数据单板 确定同步方式为更新受限的同步操作。 上述元数据单板按照确定的同步方式与申请单板进行属性同步操作包括: 当元数 据单板确定的同步方式为更新不受限的同步操作时, 元数据单板比较申请单板更新前 的属性与本地属性是否一致; 如果一致, 元数据单板向申请单板发送携带有更新操作 成功指示的响应消息, 将申请单板更新后的属性作为新属性, 并向除申请单板之外的 其它单板广播新属性; 如果不一致, 元数据单板向申请单板发送携带有更新操作失败 指示和本地属性的响应消息。 上述元数据单板按照确定的同步方式与申请单板进行属性同步操作包括: 当元数 据单板确定的同步方式为更新受限的同步操作时, 元数据单板判断是否授予申请单板 更新权限; 如果授予, 元数据单板向申请单板发送授予更新权限的指示和指示对应的 属性, 并向除申请单板之外的其它单板广播申请单板的地址和设置缓存属性为无效的 通知; 元数据单板定期接收申请单板更新后的属性, 将更新后的属性作为新属性。 上述方法还包括在以下情况之一发生时, 元数据单板取消申请单板的更新权限: 元数据单板接收到申请单板发送的属性更新完成通知; 元数据单板在设定时长内均未 收到来自申请单板的消息; 元数据单板接送到第三单板的更新权限请求, 且确认第三 单板对属性的更新优于申请单板对属性的更新。 根据本发明实施例的又一方面,提供了一种文件系统的属性同步控制装置,包括: 第一同步方式确定模块, 设置为根据当前对属性的更新操作类型确定同步方式; 其中, 该更新操作类型包括: 简易操作和复杂操作; 第一同步操作模块, 设置为按照第一同 步方式确定模块确定的同步方式与属性的元数据单板进行属性同步操作, 并在本地存 储同步后的属性。 上述第一同步方式确定模块包括: 第一同步方式确定单元, 设置为如果当前对属 性的更新操作类型为简易操作, 确定同步方式为更新不受限的同步操作; 第二同步方 式确定单元, 设置为如果当前对属性的更新操作类型为复杂操作, 确定同步方式为更 新受限的同步操作。 上述第一同步操作模块包括: 更新属性发送单元, 设置为当申请单板确定的同步 方式为更新不受限的同步操作时, 对本地的属性进行更新操作, 并将更新前的属性和 更新后的属性发送给属性的元数据单板; 第一新属性确定单元, 设置为接收元数据单 板的响应消息, 如果响应消息指示更新操作成功, 将更新后的属性作为新属性; 如果 响应消息指示更新操作失败, 将响应消息中携带的属性作为新属性。 上述第一同步操作模块包括: 权限请求单元, 设置为当申请单板确定的同步方式 为更新受限的同步操作时, 向属性的元数据单板发送获取更新权限的请求; 信息接收 单元, 设置为接收元数据单板返回的授予更新权限的指示和指示对应的属性; 第二更 新属性发送单元, 设置为对信息接收单元接收的属性进行更新操作, 并定期将更新后 的属性发送给元数据单板。 根据本发明实施例的再一方面,提供了一种文件系统的属性同步控制装置,包括: 更新请求接收模块, 设置为接收申请单板的更新请求; 第二同步方式确定模块, 设置 为根据更新请求接收模块接收的更新请求中携带的信息确定同步方式; 第二同 作 模块, 设置为按照第二同步方式确定模块确定的同步方式与申请单板进行属性同步操 作, 并在本地存储同步后的属性。 上述第二同步方式确定模块包括: 第三同步方式确定单元, 设置为如果更新请求 中携带的信息是申请单板更新前的属性和更新后的属性, 确定同步方式为更新不受限 的同步操作; 第四同步方式确定单元, 设置为如果更新请求中携带的信息是获取更新 权限的请求, 确定同步方式为更新受限的同步操作。 上述第二同步操作模块包括: 属性比较单元, 设置为当第一同步方式确定单元确 定的同步方式为更新不受限的同步操作时, 比较申请单板更新前的属性与本地属性是 否一致; 同步处理单元, 设置为如果属性比较单元的比较结果为一致, 向申请单板发 送携带有更新操作成功指示的响应消息, 将申请单板更新后的属性作为新属性, 并向 除申请单板之外的其它单板广播新属性; 如果属性比较单元的比较结果为不一致, 向 申请单板发送携带有更新操作失败指示和本地属性的响应消息。 上述第二同步操作模块包括: 权限判断单元, 设置为当第二同步方式确定单元确 定的同步方式为更新受限的同步操作时,判断是否授予申请单板更新权限; 广播单元, 设置为如果权限判断单元的判断结果为授予, 向申请单板发送授予更新权限的指示和 指示对应的属性, 并向除申请单板之外的其它单板广播申请单板的地址和设置缓存属 性为无效的通知; 属性更新单元, 设置为定期接收申请单板更新后的属性, 将更新后 的属性作为新属性。 根据本发明实施例的还一方面, 提供了一种文件系统的属性同步控制系统, 包括 申请单板和元数据单板, 其中, 申请单板包括上述本发明实施例又一方面中的属性同 步控制装置, 元数据单板包括上述本发明实施例再一方面中的属性同步控制装置。 通过本发明实施例, 按照属性的更新操作类型有针对地与元数据单板进行属性的 同步过程, 进而可以简化属性同步过程, 同时通过将同步后的属性在申请单板侧进行 存储, 为后续申请单板获取该属性提供了便捷方式, 这种方式避免了属性存储于一个 可靠节点导致系统性能下降的问题, 降低了属性的访问和修改所带来的性能和资源的 消耗, 优化了系统性能。 附图说明 此处所说明的附图用来提供对本发明实施例的进一步理解,构成本申请的一部分, 本发明的示意性实施例及其说明用于解释本发明, 并不构成对本发明的不当限定。 在 附图中: 图 1是根据本发明实施例的从申请单板侧描述的文件系统的属性同步控制方法的 流程图; 图 2是根据本发明实施例的属性缓存中单个属性的状态变迁状态图; 图 3是根据本发明实施例的从元数据单板侧描述的文件系统的属性同步控制方法 的流程图; 图 4是根据本发明实施例的简单修改属性的同步方法流程图; 图 5是根据本发明实施例的复杂修改属性的同步方法流程图; 图 6是根据本发明实施例的从申请单板侧描述的文件系统的属性同步控制装置的 结构框图; 图 Ί是根据本发明实施例的从元数据单板侧描述的文件系统的属性同步控制装置 的结构框图; 图 8是根据本发明实施例的文件系统的属性同步控制系统的结构框图。 具体实施方式 下文中将参考附图并结合实施例来详细说明本发明。 需要说明的是, 在不冲突的 情况下, 本申请中的实施例及实施例中的特征可以相互组合。 本发明实施例中的文件系统中设置有多个单板(也可以称为节点), 其中, 将存储 元数据的单板称为元数据单板, 将当前进行属性操作的单板称为申请单板, 下面分别 从申请单板和元数据单板侧描述文件系统的属性同步控制方法。 如图 1所示的文件系统的属性同步控制方法的流程图, 该方法以从申请单板侧描 述为例进行说明, 该方法包括以下步骤: 步骤 S102, 申请单板根据当前对属性的更新操作类型确定同步方式; 其中, 该更 新操作类型包括: 简易操作和复杂操作; 例如: chmod (文件权限修改)、 chown (文件属主修改)、 chgrp (文件用户组修 改)等操作均可以通过操作一次完成,这些操作就可以归类为简单操作,而像 createfile (创建文件)、 writefile (写文件) 等连续的操作中的属性不能一次更新到位, 则可以 将这些操作归类为复杂操作; 或者当前操作需要修改多个元数据的属性的情况也可以 归类为复杂操作; 步骤 S104, 申请单板按照上述确定的同步方式与属性的元数据单板进行属性同步 操作, 并在本地存储同步后的属性。 本实施例按照属性的更新操作类型有针对地与元数据单板进行属性的同步过程, 进而可以简化属性同步过程, 同时通过将同步后的属性在申请单板侧进行存储, 为后 续申请单板获取该属性提供了便捷方式, 这种方式避免了属性存储于一个可靠节点导 致系统性能下降的问题, 降低了属性的访问和修改所带来的性能和资源的消耗, 优化 了系统性能。 如果当前申请单板对属性的更新操作类型为简易操作, 则申请单板无需向元数据 单板获取更新权限, 申请单板可以确定同步方式为更新不受限的同步操作; 如果当前 对属性的更新操作类型为复杂操作, 为了避免申请单板与元数据单板间频繁的消息交 互, 本实施例的申请单板可以确定同步方式为更新受限的同步操作。 下面分别对更新 不受限的同步操作和更新受限的同步操作进行描述。 一、 更新不受限的同步操作 当申请单板确定的同步方式为更新不受限的同步操作时, 申请单板先对本地的属 性进行更新操作, 并将更新前的属性 (也可以称为原始属性) 和更新后的属性发送给 该属性的元数据单板; 申请单板接收上述元数据单板的响应消息, 如果响应消息指示更新操作成功, 则 申请单板将更新后的属性作为新属性; 如果响应消息指示更新操作失败, 申请单板将 响应消息中携带的属性作为新属性。 这种方式下, 申请单板将会在本地保存上述新属性, 当申请单板下次在需要对该 属性进行访问时, 可以直接在本地获取到该属性, 为了保证属性的有效性, 可以为该 属性设置一个状态标志位, 如果该状态标志位为 I, 则该属性为无效的属性, 否则, 该 属性有效。 二、 更新不受限的同步操作 当申请单板确定的同步方式为更新受限的同步操作时, 申请单板向该属性的元数 据单板发送获取更新权限的请求; 申请单板接收元数据单板返回的授予更新权限的指示和该指示对应的属性; 申请单板对该属性进行更新操作, 并定期将更新后的属性 (即新属性) 发送给元 数据单板。 这种同步方式之后, 申请单板将在本地存储更新后的属性, 因为这种复杂操作方 式会对属性进行多次操作, 因此本实施例采用了定期向元数据单板发送更新后的属性 的方式, 以使元数据单板上的属性为最新属性。 本发明实施例考虑到系统中有很多的单板, 为了便于其它单板就近且分散访问属 性, 本实施例使用 MOSIE (Modifier Owner Share Exclusive Invalid, 修改、 拥有、 共 享、 独享和无效) 的方式管理属性, 图 2给出了属性缓存中单个属性的状态变迁状态 图, 其中, 各个状态如下:
M: 数据的修改者位, 获得该权限的节点即拥有读写权限 (可以与其他属性位共 存, 拥有 M的节点需要记录 0的位置, 方便更新 0的状态), 相当于上述更新权限;
0: 指属性的拥有者, 只拥有只读的权限 (目前是元数据所在的地方, 同时记录 M的位置); S: 属性的共享者, 该节点只拥有属性只读的权限;
E: 属性的独享者, 拥有属性独享的权限 (此时的节点一定拥有 M权限, 却不一 定是 0, 可能 0已被置换了, 需要记录 0的位置);
I: 该状态的节点上的属性为无效的, 没有意义。 申请单板对属性进行更新操作之后, 元数据单板将会将申请单板的地址通知给其 它节点, 此后, 第三单板可以保存申请单板的地址, 在需要访问上述属性时, 向申请 单板发送请求即可。 基于此, 上述方法还包括: 申请单板接收到第三单板获取属性的 请求后, 将更新后的属性发送给第三单板。 这种方式能够使第三单板方便地获取到上 述属性, 并且可以避免大量的节点频繁地访问元数据单板,减轻了元数据单板的压力。 为了进一步降低系统将的通信量,申请单板将更新后的属性发送给第三单板之前, 上述方法还包括: 申请单板记录第三单板的地址和获取时间; 申请单板根据记录的第 三单板的获取时间判断第三单板在指定时长内的获取次数是否大于阈值, 如果是, 标 记该第三单板为该属性的热节点, 用以后续对上述属性进行更新操作之后, 将更新后 的属性直接发送给该属性的热节点。 这种方式不需要该属性的热节点发属性获取请求 给申请节点, 便可以使该属性的热节点及时地获取到更新的属性。 如图 3所示的文件系统的属性同步控制方法的流程图, 该方法以从元数据单板侧 描述为例进行说明, 该方法包括以下步骤: 步骤 S302, 元数据单板接收申请单板的更新请求; 步骤 S304, 元数据单板根据上述更新请求中携带的信息确定同步方式; 步骤 S306, 元数据单板按照确定的同步方式与上述申请单板进行属性同步操作, 并在本地存储同步后的属性。 本实施例按照申请单板的更新请求中的信息确定同步方式, 进而能够有针对地与 申请单板进行属性的同步过程, 简化了属性同步过程, 同时通过将同步后的属性在元 数据单板侧进行存储, 为后续其它单板获取该属性提供了便捷方式, 这种方式避免了 属性存储于一个可靠节点导致系统性能下降的问题, 降低了属性的访问和修改所带来 的性能和资源的消耗, 优化了系统性能。 本实施例中, 如果更新请求中携带的信息是申请单板更新前的属性和更新后的属 性, 则元数据单板可以确定同步方式为更新不受限的同步操作; 如果更新请求中携带 的信息是获取更新权限的请求,元数据单板可以确定同步方式为更新受限的同步操作。 当元数据单板确定的同步方式为更新不受限的同步操作时, 元数据单板比较上述 申请单板更新前的属性与本地属性是否一致; 如果一致, 元数据单板向申请单板发送 携带有更新操作成功指示的响应消息, 将申请单板更新后的属性作为新属性, 并向除 申请单板之外的其它单板广播该新属性; 如果不一致, 元数据单板向申请单板发送携 带有更新操作失败指示和本地属性的响应消息。这样可以简单地完成属性的更新过程。 当元数据单板确定的同步方式为更新受限的同步操作时, 元数据单板判断是否授 予申请单板更新权限 (具体判断方式可以结合当前操作的具体内容进行, 判断该申请 节点是否适合进行当前操作); 如果授予, 元数据单板向申请单板发送授予更新权限的 指示和指示对应的属性, 并向除申请单板之外的其它单板广播申请单板的地址和设置 缓存属性为无效的通知; 这种方式下, 元数据单板将可以定期接收申请单板更新后的 属性, 将更新后的属性作为新属性。 优选地, 上述方法还包括在以下情况之一发生时, 元数据单板取消申请单板的更 新权限: 1 ) 元数据单板接收到申请单板发送的属性更新完成通知; 2) 元数据单板在 设定时长内均未收到来自申请单板的消息; 3 )元数据单板接送到第三单板的更新权限 请求, 且确认第三单板对属性的更新优于申请单板对属性的更新。 本实施例通过适时 取消申请单板的更新权限, 能够使其它单板获取到该更新权限,便于属性的同步操作。 本发明上述实施例中仅对属性的更新进行了描述, 上述方法对应的系统中可以对 属性进行下述操作:
1 )获取属性: 即读取属性, 如果所读取的属性的元数据不在本板, 那么对应的元 数据的单板也会得到该属性的一份数据, 本板状态为 S, 记录元数据单板的标识 (例 如: 单板的 IP地址等可以在集群中唯一标识单板身份的一个唯一确定值), 而元数据 单板状态为 S+0, 并记录拥有的该属性的 2个单板的标识, 例如: 当一个非元数据单 板的单板 (非元数据单板) 获取了属性, 那么元数据单板需要记录本板和获取到该属 性的单板这两个单板的相关信息。
2)更新属性: 通常指修改属性, 如上述实施例所述, 本发明实施例中涉及到两种 形式: 简单模式和复杂模式, 即对应上述更新不受限的同步操作和更新受限的同步操 作。 3 ) 释放更新权限: 归还 M权限, 然后对全域做一次更新。
4)删除属性: 可以借助于广播方式通知给各个单板, 清除相关的属性缓存, 元数 据单板 (0节点) 将清除相应的元数据。 如图 4所示的简单修改属性的同步方法流程图, 其中包括以下步骤: 步骤 S402, 简单修改单板将原始属性与改后属性(即新属性) 同时发给属性的拥 有者 0 (即元数据单板); 步骤 S404, 元数据单板 (简称为 "0单板"或者 "0节点") 根据自身记录的属性与 接收到的简单修改单板发送的原始属性进行对比, 如果二者一致, 保存上述新属性, 并执行步骤 S406; 如果不一致, 执行步骤 S410; 步骤 S406, 元数据单板向简单修改单板发送修改成功通知, 然后执行步骤 S408; 步骤 S408, 元数据单板向广播域内的单板广播上述新属性; 其中, 上述步骤 S406和步骤 S408的顺序可以互换; 步骤 S410, 元数据单板向简单修改单板返回修改失败通知, 该通知携带有元数据 单板上记录的属性。 如图 5所示的复杂修改属性的同步方法流程图, 其中包括以下步骤: 步骤 S502, 复杂修改单板向元数据单板获取修改权限 M, 发送获取 M请求时, 携带有自身的原始属性; 步骤 S504,元数据单板根据自身记录的属性与接收到的复杂修改单板发送的原始 属性进行对比, 如果相同, 且复杂修改单板适合授予 M, 执行步骤 S506; 否则, 执行 步骤 S510; 步骤 S506, 元数据单板授予复杂修改单板 M, 记录复杂修改单板的地址 并将自 己记录的所有拥有本属性信息的数据返回给复杂修改单板, 然后执行步骤 S508; 步骤 S508, 元数据单板向广播域内的单板发送更新 M的通知, 该通知携带有复 杂修改单板的地址和设置缓存属性为无效的指示(Invalid); 即对非 0广播节点下发更 新 M节点信息, 同时无效该节点上的属性。 如果对方节点需要该属性, 可以向复杂修 改单板申请; 其中, 上述步骤 S506和步骤 S508的顺序可以互换; 此时复杂修改单板获得修改权限 M, 可以简称为 M节点或者 M单板, 同时复杂 修改单板将作为属性的更新方负责更新相关的修改。 具体包括:
1 ) 定期更新数据单板上的属性信息, 同时, 这种方式也是维护 M状态的心跳通 信方式, 如果连续丢失心跳(即元数据单板设定时长内为收到该心跳通信), 元数据单 板将结束复杂修改单板的 M状态,并以自己目前所得到的属性对记录的广播域内的单 板解除复杂修改单板的 M状态,并更新相关信息。同时元数据单板会将本地记录的 M 状态解除(就是集群中这个元数据失去了 M的单板, 所有的更新和读取会重新向数据 的 0 (即元数据单板) 发起请求) (本实施例中, 设置有一个保护期 30s, 在解除复杂 修改单板 M状态的 30s内, 别的单板无法申请到 M状态; 如果复杂修改单板对解除 M状态有异议, 可以再次申请 M权限, 系统将更新相关信息);
2)复杂修改单板在更新期间都没有出现异常, 复杂修改单板完成相应操作之后需 要解除 M状态, 例如: 首先对所有广播域内的单板更新属性, 同时向元数据单板发送 解除 M 状态的通知, 收到更新属性广播的单板可以给元数据单板发送确认更新的消 息, 此时元数据单板将作为同步的关键节点, 保证收到所用广播域内的消息。 步骤 S510, 元数据单板向复杂修改单板返回申请失败通知, 该通知携带有元数据 单板上记录的属性和已经具有 M权限的单板的地址(当然, 如果系统中没有单板具有 M权限, 则该通知中无此地址项)。 上述流程是复杂的属性修改流程, 目的是在文件系统创建、 写等连续操作过程中 提高本地响应速度的一种做法。 上述简单修改单板或者复杂修改单板在与其它节点进行属性同步时, 可以采用广 播、 组播或者单波的方式, 例如: 需要进行同步节点的个数超过规定个数, 则可以采 用广播模式 (需要进行同步的节点的属性状态为8, 如果没有具有 M权限的节点, 则 由 0节点进行同步), 通常系统中的一、 二层节点 (假设当前文件系统挂载在 /mnt下 面, 那么 /mnt/a这个文件或目录就是第一层节点, 而 /mnt/a/b 这个文件目录就是第二 层节点, 因为缓存的设计是树形的, 所以对于每一个缓存树, 树根部分基本上每个单 板都存在, 设计时就可以对这两层节点做特殊处理), 默认为采用广播方式。 当需要同 步属性的节点个数小于规定个数时, 可以采用组播方式进行同步; 如果需要同步属性 的节点个数为 1, 或者该节点的状态为 E (这两个节点应该是并列的状态), 则可以采 用单波方式。 上述流程中如果存在下述冲突的问题, 本实施例优选采用下述方式处理:
1、 简单修改的方式下如果有节点获得 M状态
A) 此时如果广播域内有其他单板需要进行简单修改也可以, 但是可以选择与 M 状态的节点进行通讯, 而不选择具有 0状态的节点; B) 广播域外由于获取节点信息后就进入广播域, 所以不存在广播域外简单修改 的可能。
2、 关于多个节点争抢 M状态的解决办法
A) 所有需要 M状态的操作在出现无法申请到 M状态后, 对于具有 M状态的新 节点建立简单的非 M状态操作流程; 即由复杂流程退化到简单流程, 但是操作的对象 是元数据的 M单板, 而不是 0单板, 用于当 0单板通知之前的 M单板解除 M状态 回刷数据时,该 M单板再次申请 M权限,这样说明目前的 M单板对流程的依赖很大, 此时可以由 M状态回退到 0单板上修改为新申请的 M节点退化到简单流程, 在具体 实现时, 可以通过 API程序来处理的, 能够使客户了解自己的需求去申请这个权限。 B) 对于目前的 M状态节点, 0节点将发起目前 M状态的节点解除 M状态的动 作, 以保证各个节点的公平; (即: 原始的 0节点要求收回 M权限的流程, 这个流程 后所有的 M请求单板(复杂请求)都会退化到简单请求, 该处理是不做特殊处理的一 般流程)。 3、 关于 0节点缓存不足, 需要释放掉缓存的问题 因为是缓存, 所以有可能会出现内存不够的情况, 在内存不足时, 缓存系统为了 提高性能, 一般来说采取回刷到硬盘, 下次需要时再从硬盘读取, 如果此时 0节点记 录的状态是 S (有自己以外的单板拥有这个状态), 那么在存在 M单板的时候就直接 释放掉了; 而如果没有 0节点会指定一个 S的单板为 M状态同时释放掉自己的内存, 这样控制权在 M单板上, 如果有修改的话, 由 M单板代行 0节点的权限进行处理; 当属性有修改时, M单板需要同步到 0节点; 此时 0节点再申请内存, 重新同步 M 单板修改过的属性 (此时不需要读取磁盘上数据);
0节点当发现出现 0状态必须要释放的时候, 将随机选一个节点, 并更新该节点 的状态为 M, 以保证数据可以继续运行。 对应与上述方法, 本发明实施例还提供了一种文件系统的属性同步控制装置, 该 装置可以设置在上述申请单板上, 参见图 6, 该装置包括: 第一同步方式确定模块 62, 设置为根据当前对属性的更新操作类型确定同步方 式; 其中, 更新操作类型包括: 简易操作和复杂操作; 第一同步操作模块 64, 与第一同步方式确定模块 62相连, 设置为按照第一同步 方式确定模块 62确定的同步方式与属性的元数据单板进行属性同步操作,并在本地存 储同步后的属性。 本实施例按照属性的更新操作类型有针对地与元数据单板进行属性的同步过程, 进而可以简化属性同步过程, 同时通过将同步后的属性在申请单板侧进行存储, 为后 续申请单板获取该属性提供了便捷方式, 这种方式避免了属性存储于一个可靠节点导 致系统性能下降的问题, 降低了属性的访问和修改所带来的性能和资源的消耗, 优化 了系统性能。 其中, 第一同步方式确定模块 62包括: 第一同步方式确定单元, 设置为如果当前 对属性的更新操作类型为简易操作, 确定同步方式为更新不受限的同步操作; 第二同 步方式确定单元, 设置为如果当前对属性的更新操作类型为复杂操作, 确定同步方式 为更新受限的同步操作。 优选地, 第一同步操作模块 64包括: 更新属性发送单元, 设置为当申请单板确定 的同步方式为更新不受限的同步操作时, 对本地的属性进行更新操作, 并将更新前的 属性和更新后的属性发送给属性的元数据单板; 第一新属性确定单元, 设置为接收元 数据单板的响应消息, 如果响应消息指示更新操作成功,将更新后的属性作为新属性; 如果响应消息指示更新操作失败, 将响应消息中携带的属性作为新属性。 优选地, 第一同步操作模块 64包括: 权限请求单元, 设置为当申请单板确定的同 步方式为更新受限的同步操作时, 向属性的元数据单板发送获取更新权限的请求; 信 息接收单元, 设置为接收元数据单板返回的授予更新权限的指示和指示对应的属性; 第二更新属性发送单元, 设置为对信息接收单元接收的属性进行更新操作, 并定期将 更新后的属性发送给元数据单板。 对应于上述方法, 本发明实施例还提供了一种文件系统的属性同步控制装置, 该 装置可以设置在上述元数据单板上, 参见图 7, 该装置包括: 更新请求接收模块 72, 设置为接收申请单板的更新请求; 第二同步方式确定模块 74, 与更新请求接收模块 72相连, 设置为根据更新请求 接收模块 72接收的更新请求中携带的信息确定同步方式; 第二同步操作模块 76, 与第二同步方式确定模块 74相连, 设置为按照第二同步 方式确定模块 74确定的同步方式与申请单板进行属性同步操作,并在本地存储同步后 的属性。 本实施例按照申请单板的更新请求中的信息确定同步方式, 进而能够有针对地与 申请单板进行属性的同步过程, 简化了属性同步过程, 同时通过将同步后的属性在元 数据单板侧进行存储, 为后续其它单板获取该属性提供了便捷方式, 这种方式避免了 属性存储于一个可靠节点导致系统性能下降的问题, 降低了属性的访问和修改所带来 的性能和资源的消耗, 优化了系统性能。 优选地, 第二同步方式确定模块 74包括: 第三同步方式确定单元, 设置为如果更 新请求中携带的信息是申请单板更新前的属性和更新后的属性, 确定同步方式为更新 不受限的同步操作; 第四同步方式确定单元, 设置为如果更新请求中携带的信息是获 取更新权限的请求, 确定同步方式为更新受限的同步操作。 优选地, 第二同步操作模块 76包括: 属性比较单元, 设置为当第一同步方式确定 单元确定的同步方式为更新不受限的同步操作时, 比较申请单板更新前的属性与本地 属性是否一致; 同步处理单元, 设置为如果属性比较单元的比较结果为一致, 向申请 单板发送携带有更新操作成功指示的响应消息,将申请单板更新后的属性作为新属性, 并向除申请单板之外的其它单板广播新属性;如果属性比较单元的比较结果为不一致, 向申请单板发送携带有更新操作失败指示和本地属性的响应消息。 优选地, 第二同步操作模块 76包括: 权限判断单元, 设置为当第二同步方式确定 单元确定的同步方式为更新受限的同步操作时, 判断是否授予申请单板更新权限; 广 播单元, 设置为如果权限判断单元的判断结果为授予, 向申请单板发送授予更新权限 的指示和指示对应的属性, 并向除申请单板之外的其它单板广播申请单板的地址和设 置缓存属性为无效的通知; 属性更新单元, 设置为定期接收申请单板更新后的属性, 将更新后的属性作为新属性。 对应于上述方法和装置, 本发明实施例还提供了一种文件系统的属性同步控制系 统, 图 8是根据本发明实施例的文件系统的属性同步控制系统的结构框图, 如图 8所 示,包括申请单板 60和元数据单板 70,其中, 申请单板 60包括上述图 6所示的装置, 元数据单板 70包括上述图 Ί所示的装置。 本系统可以显著提高局部访问的性能, 同时在高压力访问的情况下, 可以通过策 略, 分散系统元数据单板的处理压力。 以上实施例中的状态指一个属性对应的状态, 一个节点上可能有上百个属性, 每 个属性对应着相应的状态。 从以上的描述中可以看出, 本发明实施例使用一种高效的同步方式管理起整个文 件系统的属性, 从而使远端客户端可以根据本系统的记录数据确定出缓存的元数据是 否是可用的, 从而提高本地节点的性能, 同时又可以依托本系统, 降低属性的访问和 修改所带来的性能和资源的消耗, 进而避免了系统中的各个节点均向一个节点获取属 性引起的系统性能下降问题。 显然, 本领域的技术人员应该明白, 上述的本发明实施例的各模块或各步骤可以 用通用的计算装置来实现, 它们可以集中在单个的计算装置上, 或者分布在多个计算 装置所组成的网络上, 优选地, 它们可以用计算装置可执行的程序代码来实现, 从而, 可以将它们存储在存储装置中由计算装置来执行, 并且在某些情况下, 可以以不同于 此处的顺序执行所示出或描述的步骤, 或者将它们分别制作成各个集成电路模块, 或 者将它们中的多个模块或步骤制作成单个集成电路模块来实现。 这样, 本发明实施例 不限制于任何特定的硬件和软件结合。 以上所述仅为本发明的优选实施例而已, 并不用于限制本发明, 对于本领域的技 术人员来说, 本发明可以有各种更改和变化。 凡在本发明的精神和原则之内, 所作的 任何修改、 等同替换、 改进等, 均应包含在本发明的保护范围之内。 工业实用性 本发明实施例提供的技术方案可以应用于存储系统领域, 解决了存储于一个可靠 节点导致系统性能下降的问题, 在保证数据缓存高效性的前提下, 实现了操作更为频 繁的属性缓存的速度与性能的提高。

Claims

权 利 要 求 书
1. 一种文件系统的属性同步控制方法, 包括:
申请单板根据当前对属性的更新操作类型确定同步方式; 其中, 所述更新 操作类型包括: 简易操作和复杂操作; 所述申请单板按照确定的所述同步方式与所述属性的元数据单板进行属性 同步操作, 并在本地存储同步后的属性。
2. 根据权利要求 1所述的方法, 其中, 申请单板根据当前对属性的更新操作类型 确定同步方式包括:
如果当前对属性的更新操作类型为简易操作, 申请单板确定同步方式为更 新不受限的同步操作;
如果当前对属性的更新操作类型为复杂操作, 申请单板确定同步方式为更 新受限的同步操作。
3. 根据权利要求 2所述的方法, 其中, 所述申请单板按照确定的所述同步方式与 所述属性的元数据单板进行属性同步操作包括:
当所述申请单板确定的同步方式为更新不受限的同步操作时, 所述申请单 板对本地的属性进行更新操作, 并将更新前的属性和所述更新后的属性发送给 所述属性的元数据单板;
所述申请单板接收所述元数据单板的响应消息, 如果所述响应消息指示更 新操作成功, 所述申请单板将所述更新后的属性作为新属性; 如果所述响应消 息指示更新操作失败,所述申请单板将所述响应消息中携带的属性作为新属性。
4. 根据权利要求 2所述的方法, 其中, 所述申请单板按照确定的所述同步方式与 所述属性的元数据单板进行属性同步操作包括:
当所述申请单板确定的同步方式为更新受限的同步操作时, 所述申请单板 向所述属性的元数据单板发送获取更新权限的请求;
所述申请单板接收所述元数据单板返回的授予更新权限的指示和所述指示 对应的属性;
所述申请单板对所述属性进行更新操作, 并定期将更新后的属性发送给所 述元数据单板。
5. 根据权利要求 4所述的方法, 其中, 所述申请单板对所述属性进行更新操作之 后, 所述方法还包括:
所述申请单板接收到第三单板获取所述属性的请求后, 将更新后的属性发 送给所述第三单板。
6. 根据权利要求 5所述的方法, 其中, 所述申请单板将更新后的属性发送给所述 第三单板之前, 所述方法还包括:
所述申请单板记录所述第三单板的地址和获取时间;
所述申请单板根据记录的所述第三单板的获取时间判断所述第三单板在指 定时长内的获取次数是否大于阈值, 如果是, 标记所述第三单板为所述属性的 热节点, 用以后续对所述属性进行更新操作之后, 将更新后的所述属性发送给 所述属性的热节点。
7. 一种文件系统的属性同步控制方法, 包括:
元数据单板接收申请单板的更新请求;
所述元数据单板根据所述更新请求中携带的信息确定同步方式; 所述元数据单板按照确定的所述同步方式与所述申请单板进行属性同步操 作, 并在本地存储同步后的属性。
8. 根据权利要求 7所述的方法, 其中, 所述元数据单板根据所述更新请求中携带 的信息确定同步方式包括:
如果所述更新请求中携带的信息是所述申请单板更新前的属性和更新后的 属性, 所述元数据单板确定同步方式为更新不受限的同步操作;
如果所述更新请求中携带的信息是获取更新权限的请求, 所述元数据单板 确定同步方式为更新受限的同步操作。
9. 根据权利要求 8所述的方法, 其中, 所述元数据单板按照确定的所述同步方式 与所述申请单板进行属性同步操作包括:
当所述元数据单板确定的同步方式为更新不受限的同步操作时, 所述元数 据单板比较所述申请单板更新前的属性与本地属性是否一致;
如果一致, 所述元数据单板向所述申请单板发送携带有更新操作成功指示 的响应消息, 将所述申请单板更新后的属性作为新属性, 并向除所述申请单板 之外的其它单板广播所述新属性; 如果不一致, 所述元数据单板向所述申请单板发送携带有更新操作失败指 示和所述本地属性的响应消息。
10. 根据权利要求 8所述的方法, 其中, 所述元数据单板按照确定的所述同步方式 与所述申请单板进行属性同步操作包括:
当所述元数据单板确定的同步方式为更新受限的同步操作时, 所述元数据 单板判断是否授予所述申请单板更新权限;
如果授予, 所述元数据单板向所述申请单板发送授予更新权限的指示和所 述指示对应的属性, 并向除所述申请单板之外的其它单板广播所述申请单板的 地址和设置缓存属性为无效的通知; 所述元数据单板定期接收所述申请单板更 新后的属性, 将所述更新后的属性作为新属性。
11. 根据权利要求 10所述的方法, 其中, 所述方法还包括在以下情况之一发生时, 所述元数据单板取消所述申请单板的更新权限:
所述元数据单板接收到所述申请单板发送的属性更新完成通知; 所述元数据单板在设定时长内均未收到来自所述申请单板的消息; 所述元数据单板接送到第三单板的更新权限请求, 且确认所述第三单板对 所述属性的更新优于所述申请单板对所述属性的更新。
12. 一种文件系统的属性同步控制装置, 包括:
第一同步方式确定模块, 设置为根据当前对属性的更新操作类型确定同步 方式; 其中, 所述更新操作类型包括: 简易操作和复杂操作;
第一同步操作模块, 设置为按照所述第一同步方式确定模块确定的所述同 步方式与所述属性的元数据单板进行属性同步操作, 并在本地存储同步后的属 性。
13. 根据权利要求 12所述的装置, 其中, 所述第一同步方式确定模块包括:
第一同步方式确定单元, 设置为如果当前对属性的更新操作类型为简易操 作, 确定同步方式为更新不受限的同步操作;
第二同步方式确定单元, 设置为如果当前对属性的更新操作类型为复杂操 作, 确定同步方式为更新受限的同步操作。
14. 根据权利要求 13所述的装置, 其中, 所述第一同步操作模块包括: 更新属性发送单元, 设置为当所述申请单板确定的同步方式为更新不受限 的同步操作时, 对本地的属性进行更新操作, 并将更新前的属性和所述更新后 的属性发送给所述属性的元数据单板;
第一新属性确定单元, 设置为接收所述元数据单板的响应消息, 如果所述 响应消息指示更新操作成功, 将所述更新后的属性作为新属性; 如果所述响应 消息指示更新操作失败, 将所述响应消息中携带的属性作为新属性。
15. 根据权利要求 13所述的装置, 其中, 所述第一同步操作模块包括:
权限请求单元, 设置为当所述申请单板确定的同步方式为更新受限的同步 操作时, 向所述属性的元数据单板发送获取更新权限的请求;
信息接收单元, 设置为接收所述元数据单板返回的授予更新权限的指示和 所述指示对应的属性;
第二更新属性发送单元, 设置为对所述信息接收单元接收的所述属性进行 更新操作, 并定期将更新后的属性发送给所述元数据单板。
16. 一种文件系统的属性同步控制装置, 包括:
更新请求接收模块, 设置为接收申请单板的更新请求;
第二同步方式确定模块, 设置为根据所述更新请求接收模块接收的所述更 新请求中携带的信息确定同步方式;
第二同步操作模块, 设置为按照所述第二同步方式确定模块确定的所述同 步方式与所述申请单板进行属性同步操作, 并在本地存储同步后的属性。
17. 根据权利要求 16所述的装置, 其中, 所述第二同步方式确定模块包括: 第三同步方式确定单元, 设置为如果所述更新请求中携带的信息是所述申 请单板更新前的属性和更新后的属性,确定同步方式为更新不受限的同步操作; 第四同步方式确定单元, 设置为如果所述更新请求中携带的信息是获取更 新权限的请求, 确定同步方式为更新受限的同步操作。
18. 根据权利要求 17所述的装置, 其中, 所述第二同步操作模块包括:
属性比较单元, 设置为当所述第一同步方式确定单元确定的同步方式为更 新不受限的同步操作时,比较所述申请单板更新前的属性与本地属性是否一致; 同步处理单元, 设置为如果所述属性比较单元的比较结果为一致, 向所述 申请单板发送携带有更新操作成功指示的响应消息, 将所述申请单板 ^ S ^ 属性作为新属性, 并向除所述申请单板之外的其它单板广播所述新属性; 如果 所述属性比较单元的比较结果为不一致, 向所述申请单板发送携带有更新操作 失败指示和所述本地属性的响应消息。
19. 根据权利要求 17所述的装置, 其中, 所述第二同步操作模块包括:
权限判断单元, 设置为当所述第二同步方式确定单元确定的同步方式为更 新受限的同步操作时, 判断是否授予所述申请单板更新权限;
广播单元, 设置为如果所述权限判断单元的判断结果为授予, 向所述申请 单板发送授予更新权限的指示和所述指示对应的属性, 并向除所述申请单板之 外的其它单板广播所述申请单板的地址和设置缓存属性为无效的通知;
属性更新单元, 设置为定期接收所述申请单板更新后的属性, 将所述更新 后的属性作为新属性。
20. 一种文件系统的属性同步控制系统, 包括申请单板和元数据单板, 其中, 所述 申请单板包括权利要求 12-15中任一项所述的装置, 所述元数据单板包括权利 要求 16-19中任一项所述的装置。
PCT/CN2013/082289 2013-03-08 2013-08-26 文件系统的属性同步控制方法、装置和系统 WO2014134905A1 (zh)

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN201310074099.5 2013-03-08
CN201310074099.5A CN104035944B (zh) 2013-03-08 2013-03-08 文件系统的属性同步控制方法、装置和系统

Publications (1)

Publication Number Publication Date
WO2014134905A1 true WO2014134905A1 (zh) 2014-09-12

Family

ID=51466715

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2013/082289 WO2014134905A1 (zh) 2013-03-08 2013-08-26 文件系统的属性同步控制方法、装置和系统

Country Status (2)

Country Link
CN (1) CN104035944B (zh)
WO (1) WO2014134905A1 (zh)

Families Citing this family (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN111638897B (zh) * 2020-06-02 2023-09-26 驭新智行科技(宁波)有限公司 系统更新方法、装置、终端设备及存储介质

Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN101385030A (zh) * 2006-02-15 2009-03-11 微软公司 同步对等参与者模型
CN102339283A (zh) * 2010-07-20 2012-02-01 中兴通讯股份有限公司 集群文件系统访问控制方法及集群节点
CN102521297A (zh) * 2011-11-30 2012-06-27 北京人大金仓信息技术股份有限公司 无共享数据库集群中实现系统动态扩展的方法

Family Cites Families (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20020199017A1 (en) * 2001-06-25 2002-12-26 Russell Lance W. Routing meta data for network file access

Patent Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN101385030A (zh) * 2006-02-15 2009-03-11 微软公司 同步对等参与者模型
CN102339283A (zh) * 2010-07-20 2012-02-01 中兴通讯股份有限公司 集群文件系统访问控制方法及集群节点
CN102521297A (zh) * 2011-11-30 2012-06-27 北京人大金仓信息技术股份有限公司 无共享数据库集群中实现系统动态扩展的方法

Also Published As

Publication number Publication date
CN104035944A (zh) 2014-09-10
CN104035944B (zh) 2018-11-09

Similar Documents

Publication Publication Date Title
CN104715001B (zh) 用于对数据处理系统的集群中的共享资源执行写入操作的方法和系统
US10374798B2 (en) System and method for rotating client security keys
EP3356961B1 (en) Peer-to-peer syncable storage system
US9880753B2 (en) Write requests in a distributed storage system
US7007047B2 (en) Internally consistent file system image in distributed object-based data storage
US8935560B2 (en) System and method of file locking in a network file system federated namespace
JP5028516B2 (ja) 分散型メッシュネットワーク
US11399014B2 (en) System and method of obtaining data from private cloud behind enterprise firewall
AU2015410318C1 (en) Data write method, apparatus, and system
US10846185B2 (en) Method for processing acquire lock request and server
EP3127018A1 (en) Geographically-distributed file system using coordinated namespace replication
JP2005535947A (ja) 異なるタイプのバックエンド・データ・ストアにアクセスするためのシステムおよび方法
JP2006221637A (ja) ネットワーク内のコンテンツに対するユーザーアクセス制御システム及び方法
JP2015535970A5 (zh)
TW201735586A (zh) 路由表的同步方法、裝置及系統
CN104360981B (zh) 面向多核多处理器平台的Cache一致性协议的设计方法
CN106612339A (zh) 域名更新方法、系统和主dns服务器
WO2013174195A1 (zh) 文档权限控制方法、装置和系统
JP2018515850A (ja) データ・コピー方法及びデバイス
US9871863B2 (en) Managing network attached storage
CN109951567A (zh) 一种双数据中心应用部署方法
CN104021137A (zh) 一种基于目录授权的客户端本地开闭文件的方法及系统
WO2014134905A1 (zh) 文件系统的属性同步控制方法、装置和系统
JP7250121B2 (ja) クラウドデグレードモードにおいてデバイス動作信頼性を継続的に確保するための方法および装置
CN114880717A (zh) 数据归档方法及装置

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 13877391

Country of ref document: EP

Kind code of ref document: A1

NENP Non-entry into the national phase

Ref country code: DE

122 Ep: pct application non-entry in european phase

Ref document number: 13877391

Country of ref document: EP

Kind code of ref document: A1