JP5257444B2 - Meta server and file management system - Google Patents

Description

  The present invention relates to a meta server, and more particularly to a meta server that manages mapping information for managing allocation of data constituting a file to a storage apparatus. The present invention relates to a file management system, a file management method, and a program for managing files in a file system using such a metaserver.

  In a file system, there is a file system in a form in which the configuration of data blocks or objects constituting a file is not managed by a host or storage device but is managed using a device (metaserver) that performs block allocation management (for example, non-file system). Patent Document 1). FIG. 13 shows a file system having such a configuration. The meta server 103 stores mapping information indicating to which area (block) in the storage device the data block is mapped for each data block constituting the file.

  FIG. 14 shows a sequence when accessing the meta server. At the time of access such as read / write, the host 101 inquires of the meta server 103 about mapping information of the data block in the file to be accessed (step S101). The meta server 103 returns the mapping information of the corresponding data block to the host 101 (step S102). The host 101 accesses the storage apparatus 102 using the acquired mapping information (step S103).

Literature list:
IPSJ 57th (late 1998) national convention performance proceedings (1) "Overview of parallel file system PF3" Junichi Yamato, Hiroyuki Otani, Yuichi Aiba, Hisayuki Aoki NEC C & C Media Research Labs p99, p100, and "File management method in parallel file system PF3" Hiroyuki Otani, Junichi Yamato, Yuichi Aiba, Hisayuki Aoki NEC C & C Media Research Labs p101, p102, Publication date: October 5, 1998

  In the file system using the meta server, it is necessary to inquire the meta server 103 for mapping information every time the host 101 accesses the storage apparatus 102. For this reason, the response time of file access becomes long. A method in which the host 101 caches the mapping information acquired in step S102 and shortens the response time is also conceivable. However, even in that case, the response time cannot be shortened except when the same data block is repeatedly accessed.

  It is an object of the present invention to provide a file management system, a file management method, and a program that can be used in a file system having an allocation management device (metaserver) and can reduce response time during access.

  The present invention is a file management method used in a file system having a meta server for managing mapping information for managing allocation of data constituting a file to a storage device, and when the meta server receives a file open request from a host, Provided is a file management method comprising: acquiring mapping information of a file to be opened from a mapping information storage unit for storing mapping information; and the metaserver transmitting the acquired mapping information to the host. To do.

  The present invention is also a metaserver for managing mapping information for managing allocation of data constituting a file to a storage device, and when there is a file open request from a host, from a mapping information storage unit for storing the mapping information, There is provided a meta server comprising: a mapping information acquisition unit that acquires mapping information of a file to be opened; and a mapping information transmission unit that transmits the mapping information acquired by the mapping information acquisition unit to the host.

  The present invention is further a meta server for managing mapping information for managing allocation of data constituting a file to a storage device, and when there is a file open request from a host, from a mapping information storage unit for storing the mapping information, A meta server having a mapping information acquisition unit that acquires mapping information of a file to be opened, and a mapping information transmission unit that transmits mapping information acquired by the mapping information acquisition unit to the host, and is transmitted from the meta server when the file is opened. And a host having a mapping information cache for registering the mapping information in the cache.

  The present invention is also a program for a file system having a meta server that manages mapping information for managing allocation of data constituting a file to a storage device, and when the meta server receives a file open request from a host, There is provided a program characterized by executing a process of acquiring mapping information of a file to be opened from a mapping information storage unit for storing mapping information and a process of transmitting the acquired mapping information to the host.

  The file management system, file management method, metaserver, and program of the present invention can shorten the response time when the host accesses data.

  The above and other objects, features, and advantages of the present invention will become apparent from the following description with reference to the drawings.

The block diagram which shows the file management system of 1st Embodiment of this invention. (A)-(d) is a diagram which shows the example of data of mapping information. (A)-(d) is a diagram which shows another example of data of mapping information. The flowchart which shows the process sequence of the host at the time of file opening. The flowchart which shows the process sequence of the metaserver at the time of file opening. The block diagram which illustrates the rule used in the case of mapping information acquisition range determination. The flowchart which shows the process sequence of mapping information acquisition. (A)-(d) is a diagram which shows the example of data of a mapping information cache. (A)-(d) is a diagram which shows another example of data of a mapping information cache. The flowchart which shows the process sequence of the host at the time of file access. The flowchart which shows the process sequence of the metaserver at the time of file access. The block diagram which shows the file management system of 2nd Embodiment of this invention. The block diagram which shows the file system using a metaserver. FIG. 14 is a sequence diagram showing a flow of processing when accessing a file in the file system shown in FIG. 13.

  Hereinafter, embodiments of the present invention will be described in detail with reference to the drawings. FIG. 1 shows a file management system according to a first embodiment of the present invention. The file management system includes a host 10, a storage device 20, and a meta server 30. The host 10 is a computer on which a user program operates. The storage device 20 is a storage device that stores data. The meta server 30 is a computer that manages mapping information that is a correspondence relationship between the data blocks constituting the file and the data blocks in the storage apparatus 20. The functions of each unit in the host 10 and the meta server 30 can be realized by executing a predetermined program on the computer. Although only one host 10 is shown in FIG. 1, a plurality of hosts 10 may be provided.

  The host 10 has a file access unit 11. A user program is running on the host 10. The host 10 requests the meta server 30 to open the file prior to access (read / write) to the file. After the file open, the user program accesses the data of the target file in the storage apparatus 20 based on the mapping information managed by the meta server 30 via the file access unit 11.

  The file access unit 11 includes a mapping information cache 12 that caches the mapping information acquired from the meta server 30. The mapping information cache 12 is a storage device that holds mapping information. When the mapping information corresponding to the data block to be accessed exists in the mapping information cache 12, the file access unit 11 accesses the data block using the mapping information. When there is no mapping information corresponding to the data block to be accessed in the mapping information cache 12, the file access unit 11 acquires the mapping information of the data block to be accessed from the meta server 30.

  The meta server 30 includes a file open request reception unit 31, a mapping information acquisition request reception unit 32, an open mapping information return range determination unit 33, a mapping information acquisition unit 34, a response return unit 35, and a mapping information storage unit 36. The mapping information storage unit 36 stores mapping information in the storage device. The mapping information acquisition request receiving unit 32 receives a mapping information acquisition request from the host. When receiving the mapping information acquisition request, the mapping information acquisition request receiving unit 32 instructs the mapping information acquisition unit 34 to acquire the mapping information corresponding to the file and data block specified in the request. The mapping information acquisition unit 34 acquires mapping information indicating in which area in the storage apparatus 20 the data block to be accessed is stored from the mapping information storage unit 36 and passes it to the response return unit 35. The response return unit 35 returns a response to the mapping information acquisition including the mapping information acquired by the mapping information acquisition unit 34 to the host 10.

  The file open request receiving unit 31 receives a file open request from the host 10. When receiving the file open request, the file open request receiving unit 31 instructs the open mapping information return range determining unit 33 to determine the range of mapping information to be returned to the host 10 together with the file open request. Upon receiving the range determination instruction, the open mapping information return range determination unit 33 determines the range of mapping information to be returned to the host 10 together with a response to the file open. The mapping information return range determination unit 33 at the time of file open determines, for example, mapping information corresponding to the range of the data area that is expected to be accessed after the file open as the range of mapping information to be returned. The open mapping information return range determination unit 33 instructs the mapping information acquisition unit 34 to acquire mapping information in the determined range after the range is determined. The mapping information acquisition unit 34 acquires the mapping information in the specified range from the mapping information storage unit 36 and passes it to the response return unit 35. The response return unit 35 includes the mapping information acquired by the mapping information acquisition unit 34 in the response to the file open request and returns it to the host 10.

  The storage device 20 is a magnetic disk, a magneto-optical disk, an optical disk, a nonvolatile memory, and an array device thereof that store data. There may be a plurality of storage devices 20. If there are a plurality, the mapping information includes the identifier of the storage device that stores the data block. In this case, the host 10 refers to the mapping information and accesses the corresponding storage apparatus 20 based on the identifier of the storage apparatus. The storage device 20 may be a device such as an OSD (Object-Based Storage Device). In the present embodiment, the storage apparatus 20 is mainly described as a block access apparatus. However, the present invention is applicable to OSD by handling data blocks as objects.

  The meta server 30 manages the mapping information as a table composed of mapping information records. 2A to 2D show examples of records stored in the mapping information table. The mapping information record includes area information in the file and information on the storage device in which the area is actually stored. FIGS. 2A and 2B are examples when the storage apparatus is a block device. The area information in the file is composed of a set of the start address and end address of the area in the file or a set of the start address and the area size. The area information in the storage apparatus is composed of a set of the start address and end address of the area in the storage apparatus or a set of the start address and area size. FIGS. 2C and 2D are examples when the storage device is an OSD. In the case of OSD, the object ID is information of the storage device that is stored in practice.

  FIGS. 3A to 3D show other examples of records stored in the mapping information table. These are examples when the file is divided and stored in a plurality of storage devices. 3A and 3B are examples in the case where the storage apparatus is a block device. The difference from FIGS. 2A and 2B is that information (storage device ID) for specifying the storage device storing the area is added. FIGS. 3C and 3D are examples when the storage device is an OSD. A difference from FIGS. 2C and 2D is that a storage apparatus ID is added. As the storage device ID, a storage device ID, SCSI ID, Fiber Channel WWN (world wide name), IP address, MAC address, URL, or the like can be used.

  FIG. 4 shows a processing procedure of the host 10 when the file is open. When opening a file, the path name of the file is designated to the host 10 by an operator or the like. First, the host 10 acquires file identification information for acquiring mapping information from the path name of the file (step A1). Step A1 is processing performed in another server or directory server, and is processing for acquiring file identification information such as a file handle for acquiring file mapping information (path name resolution processing). The path name resolution process is the same as the path name resolution by lookup in NFS, and details thereof are omitted.

  Next, the host 10 designates the file identifier to the meta server 30 and requests file open processing (step A2), and waits until a response is returned (step A3). When the meta server 30 receives the open request, the meta server 30 performs processing to open the file. FIG. 5 shows an open process procedure in the metaserver 30. The meta server 30 receives the open request from the file open request receiving unit 31. The file open request receiving unit 31 searches for information on the file corresponding to the file identifier specified in the open request (step B1). The meta server 30 determines whether or not the corresponding file information is found as a result of the search, that is, whether or not the mapping information for the data block constituting the corresponding file exists in the mapping information storage unit 36 ( Step B2).

  The meta server 30 manages the correspondence between the files and the data blocks constituting the file, and the correspondence between each data block and the mapping information record. The file open request receiving unit 31 can determine whether or not the mapping information for the data block constituting the corresponding file exists in the mapping information storage unit 36 by referring to this correspondence. When the file open request reception unit 31 determines that the mapping information exists, the file open request reception unit 31 sets the corresponding file as a file to be opened.

  When determining that there is no mapping information in step B2, the file open request receiving unit 31 determines whether file creation is specified in the file open request (step B3). If the file open request accepting unit 31 determines that file creation is not designated in step B3, the file open request accepting unit 31 instructs the response return unit 35 to return an error response indicating that the open has failed. The response return unit 35 returns an error response to the host 10 in accordance with this instruction (step B5), and ends the process. When determining that file creation is specified in step B3, the file open request receiving unit 31 creates and initializes a new area for the file (step B4). The file open request receiving unit 31 sets the file created in step B4 as a file to be opened.

  The file open request receiving unit 31 changes the status of the open process target file, that is, the file searched in step B1 or the file created in step B4 to the open state (step B6). Thereafter, the open mapping information return range determination unit 33 is instructed to determine the range of mapping information included in the response to be returned to the host 10. Upon receiving this instruction, the open mapping information return range determination unit 33 determines the range of the mapping information (step B7). The open mapping information return range determination unit 33 notifies the mapping information acquisition unit 34 of the range determined in step B7 and requests acquisition of mapping information in that range.

  The range determination performed by the open mapping information return range determination unit 33 will be described. The open mapping information return range determination unit 33 can determine mapping information corresponding to a data area of a predetermined size determined in advance from the top of a file to be opened as a range of mapping information included in the response. . Further, the open mapping information return range determination unit 33 can determine mapping information corresponding to a data area of a predetermined size from the end of the file as a mapping information range included in the response. The open mapping information return range determination unit 33 includes mapping information corresponding to both a data area of a predetermined size from the beginning of the file and a data area of the predetermined size counted from the end of the file. It can also be determined as a range.

  The open mapping information return range determination unit 33 determines a range of mapping information to be returned using a rule that specifies mapping information corresponding to which data area in the file is mapping information included in the response. You can also. The rule may be common to each file or may be prepared for each file. When preparing for each file, it is only necessary to set file information and rules, and apply only when the file information matches. In this case, it is possible to set a rule for each file. Alternatively, the file meta information may have an attribute indicating the file type, the file type and the rule may be set as a set, and a rule having the same type may be applied. In this case, in the case of a database table file, a database index file, a moving image file, and a moving image file, an appropriate rule can be set according to the file type such as an encoding method.

  An example of the rule is shown in FIG. Rules are created in advance by a system administrator or the like. Or you may create arbitrarily. In FIG. 6, the file information entry stores information for specifying a file to which this rule is applied. As the file information, a file name, a file extension, or the like can be used. In this example, n data areas are designated from area 1 to area n. Arbitrary areas are set in the areas 1 to n. These data areas do not have to be continuous data areas. One area may be used. The area specified by the rule is preferably a data area that is expected to be accessed after the file open according to the data characteristics of the file. When a plurality of data areas are specified as in the rule shown in FIG. 6, the open mapping information return range determination unit 33 sets a range group composed of a combination of the start address and end address of each data area as mapping information. The data is passed to the acquisition unit 34.

  In FIG. 6, the area is specified by a combination of a start address and an end address, but the area may be specified by a combination of a start address and the size of the area. An area may be specified by a combination of an end address and an area size. In the rule, the designation of the area by the combination of the start address and the area size and the designation of the area by the combination of the end address and the area size may be mixed. In this case, a flag indicating whether the area is specified from the end address or the area specified from the end address may be provided in the specification of each area. By specifying an area based on the start or end address and the area size, an area within a predetermined range from the start or end can be specified by a rule for files of various lengths.

  Returning to FIG. 5, the mapping information acquisition unit 34 acquires the mapping information corresponding to the area designated from the open mapping information return range determination unit 33 (step B <b> 8). If the range determined by the file open mapping information return range determination unit 33 is a range that spans a plurality of data blocks, the mapping information acquisition unit 34 acquires a plurality of mapping information records (FIG. 2). The mapping information acquisition unit 34 passes the acquired mapping information to the response return unit 35. The response return unit 35 returns the successful response to the file open request including the mapping information acquired in Step B8 to the host 10 (Step B9).

  FIG. 7 shows a processing procedure of the mapping information acquisition unit 34. The mapping information acquisition unit 34 receives the target file and the data range in the target file, and acquires mapping information. The mapping information acquisition unit 34 selects a data range to be acquired (step C1). In Step C1, when a plurality of areas are passed to the mapping information acquisition unit 34, one area is selected from the area group. The mapping information acquisition unit 34 sets the start address of the area selected in step C1 as a search target address (step C2).

  The mapping information acquisition unit 34 searches the mapping information of the designated file for a mapping information entry that includes the search target address set in step C2 in the area information range (step C3). The mapping information acquisition unit 34 creates mapping information to be returned to the host 10 based on the mapping information entry searched in step C3 (step C4). In step C4, the mapping information acquisition unit 34 acquires the content of the mapping information entry searched in step C3 from the mapping information storage unit 36, and adds the acquired content to the mapping information to be returned to the host.

  The mapping information acquisition unit 34 determines whether or not the region information of the mapping information entry includes the end address of the data range selected in Step C1 (Step C5). When the end address is not included in the data area range of the mapping information entry, the next address after the end address of the mapping information entry searched in step C3 is obtained as the search target address in order to obtain the mapping information of the subsequent data range. (Step C6), the process returns to Step C3.

  When the mapping information acquisition unit 34 determines in step C5 that the end address of the data range is included in the data area range of the mapping information entry, that is, when mapping information for all of the data range selected in step C1 is obtained, The mapping information acquisition for the data range ends. The mapping information acquisition unit 34 determines whether there is a data range for which mapping information has not yet been acquired (step C7). If it exists, the mapping information acquisition unit 34 selects one from the unprocessed data range (step C8), returns to step C2, and acquires the mapping information of the selected data range.

  If the mapping information acquisition unit 34 determines that there is no unprocessed data range in step C7, it passes the created mapping information to the response return unit 35 (step C9) and ends the process. The mapping information created by the mapping information acquisition unit 34 is transmitted to the host 10 by the response return unit 35 in step B9 of FIG.

  If there is no corresponding mapping information entry in step C3, the area has not been allocated to the storage apparatus 20 yet. In this case, the mapping information acquisition unit 34 notifies that the allocation is not performed if the mapping information is acquired at the time of the file open, shifts the search address later, and indicates the allocated area existing up to the end of the area. Get mapping information.

  Returning to FIG. 4, when receiving the response to the open request, the host 10 registers the mapping information included in the response in the mapping information cache 12 in association with the file information (step A4). Note that step A2 and step A3 are not necessary when only path name resolution is performed and there is no explicit open processing, as in NFSv2. In that case, mapping information is returned during the path name resolution processing in step A1, and therefore, in step A4, the mapping information returned at that time may be registered in the mapping information cache 12.

  8A to 8D show examples of mapping information records recorded in the mapping information cache 12. FIGS. 9A to 9D show another example of mapping information records recorded in the mapping information cache 12. The mapping information record recorded in the mapping information cache 12 specifies the mapping information (FIGS. 2A to 2D or FIGS. 3A to 3D) in the meta server 30 and the file. File information. The file information includes a combination of information specifying a file system and an inode number or a vnode number of the file system, a combination of information specifying an NFS server in NFS (Network File System) and a file handle, or a file Can be used.

  Subsequently, the host 10 performs an open process of the storage apparatus 20 (step A5). The open process in the storage apparatus in step A5 is a process for acquiring information for accessing the storage apparatus. This open process is the same process as the LOGIN process in iSCSI, and the details are omitted. Note that step A5 may be omitted if the storage apparatus 20 does not require open processing.

  FIG. 10 shows a processing procedure when a file is accessed (read / write) from the host 10. The host 10 instructs the file access unit 11 to specify a file to be accessed and an area to be accessed and to access the file. The area to be accessed is a range of a data block, and is specified by a pair of a head address and a terminal address or a pair of a head address and a size.

  When an access to a file occurs from a user program or the like, the file access unit 11 searches the mapping information cache 12 (step D1), and whether there is mapping information corresponding to the access target data area for the access target file. It is determined whether or not (step D2). When mapping information corresponding to the entire range of the data area to be accessed exists in the mapping information cache 12, the necessary mapping information is acquired from the mapping information cache 12, and the process proceeds to access to the storage apparatus 20.

  If the file access unit 11 determines in step D2 that there is no mapping information for the entire access range or there is no mapping information for a part of the area to be accessed, the file access unit 11 transmits the mapping information for the nonexistent area. And requests the meta server 30 (step D3), and waits for a response from the meta server 30 (step D4).

  The host 10 notifies the meta server 30 of “file information” for which mapping information is to be acquired and one or more “information indicating the data range from which the mapping information is acquired”, and requests transmission of the mapping information. The file information is information for specifying a file. For the file information, a combination of information specifying the file system and the inode number or vnode number of the file system, a combination of information specifying the NFS server in NFS and the file handle, a path name of the file, etc. it can. The data range for acquiring the mapping information is designated by a set of the start address of the area in the file and the end address of the area in the file, or a set of the start address of the area in the file and the size of the area.

  FIG. 11 shows a processing procedure of the meta server 30 when there is a mapping information acquisition request from the host 10. The meta server 30 receives the mapping information acquisition request transmitted from the host 10 by the mapping information acquisition request receiving unit 32. The mapping information acquisition request reception unit 32 instructs the mapping information acquisition unit 34 to acquire mapping information with the specified file mapping information as a search target and the specified data area as a search range (step E1). The mapping information acquisition unit 34 acquires mapping information in accordance with an instruction from the mapping information acquisition request reception unit 32 (step E2). The mapping information acquisition by the mapping information acquisition unit 34 at this time is the same as the procedure shown in FIG.

  If there is no corresponding mapping information entry in step C3 of FIG. 7, the area has not been allocated to the storage apparatus 20 yet. In this case, if the mapping information is acquired with read, the allocated area is notified until the end of the area by notifying that it is not allocated, shifting the search address, as in the open process. Get mapping information. In the case of write, the data block or object in the storage apparatus is allocated. This allocation process is the same as the data block allocation in the file system. In the case of OSD, an unused object ID is assigned. When the storage device 20 is composed of a plurality of storage devices, the storage device to which the area is assigned is also determined before the entry in the storage device is assigned by write.

  In step E2, the mapping information acquisition unit 34 acquires, from the mapping information storage unit 36, mapping information corresponding to the data area accessed by the host 10 and accessed by the file to be accessed. The response return unit 35 receives the mapping information from the mapping information acquisition unit 34, and returns a response including the received mapping information to the host 10 (step E3).

  Returning to FIG. 10, when the file access unit 11 receives the mapping information from the meta server 30, the file access unit 11 registers the received mapping information in the mapping information cache 12 in association with the file information of the corresponding file (step D5). The mapping information cache registration is the same as the operation of the data cache operating on the host 10 or the NFS DN (Directory Name Lookup Cache) except that the information to be registered is mapping information. The file access unit 11 accesses the storage device 20 using the mapping information searched in step D1 or the mapping information acquired from the meta server 30 in step D3 (step D6), and ends the process.

  The read / write access processing in the storage device 20 is the same as the read / write processing in the disk device, NAS (Network Attached Storage), OSD, disk array device, and the like. In FIG. 10, after the mapping information is registered in the cache in step D5, the storage apparatus 20 is accessed in step D6. Either of these processes may be performed first. Alternatively, access to the storage device 20 and cache registration of mapping information may be performed in parallel.

  In the present embodiment, the meta server 30 returns mapping information of the file to be opened to the host 10 at the time of the file open request before the host 10 actually reads / writes the file. The host 10 registers the mapping information transmitted from the meta server 30 when the file is open in the mapping information cache 12. When the host 10 accesses the file after the file open, if the mapping information corresponding to the data area to be accessed exists in the mapping information cache 12, the host 10 can access the data area using the mapping information. . That is, the host 10 can access the file without requesting the meta server 30 to acquire the mapping information again by previously registering the mapping information in the cache after the file open. By sending the mapping information of the data area expected to be accessed after the file open from the meta server 30 to the host 10 at the time of the file open, the host 10 can immediately access the data area, Access response time can be shortened.

  The file access unit 11 also registers the mapping information acquired from the meta server 30 at the time of file access in the mapping information cache 12 (step D5 in FIG. 10). By using this mapping information cache, it is possible to shorten the access response time when accessing the same data area of the same file again. However, with only the mapping information cache at the time of file access, it is not possible to shorten the access response time for a file that has not been accessed just by opening the file. In the present embodiment, when the file is opened, mapping information that is not necessary in the file open is transmitted from the meta server 30 to the host 10 and is registered in the mapping information cache 12 so that an access response after the file is opened. Time can be reduced.

  The meta server 30 in the file management system of the above embodiment includes a file open request reception unit 31, a mapping information acquisition unit 34, and a mapping information transmission unit (response return unit) 35 as the basic configuration of the meta server of the present invention. When there is a file open request from the host 10, the mapping information acquisition unit 34 acquires the mapping information of the file to be opened from the mapping information storage unit 36 that stores the mapping information. The mapping information transmission unit (response return unit) 35 transmits the mapping information acquired by the mapping information acquisition unit 34 to the host 10 that transmitted the file open request via the communication line. The mapping information storage unit 36 may be connected to the outside of the meta server 30 via the communication line 40 with the meta server 30.

  In the file management system of the above embodiment, and the meta server, file management method, and program used in the system, the host 10 can acquire mapping information from the meta server 30 when a file open request is transmitted. Can reduce the access time required to access files.

  In the above-described embodiment, a configuration in which mapping information is cached is adopted. Since the mapping information has a smaller capacity than the data itself, there is an advantage that the storage device 20 of the host 10 only needs a small amount of cache capacity (storage capacity). In addition, the amount of communication in communication related to mapping information acquisition can be reduced. As a form for transmitting mapping information from the meta server 30 to the host 10 at the time of opening, a form in which mapping information is transmitted separately from a response to an open request triggered by open is also conceivable. However, in this case, the number of communications between the host 10 and the meta server 30 increases. From the viewpoint of reducing the number of communications, it is advantageous to transmit the response including the mapping information in the response to the open request. is there.

  As an operation of the mapping information cache 12, an operation of invalidating the cache (deleting mapping information) at regular intervals can be considered. In such a case, it is preferable that the range of mapping information transmitted from the meta server 30 to the host 10 when the file is open corresponds to a data area that may be accessed within the invalidation time. The data area that may be accessed within the invalidation time is determined by the system administrator or the like based on the access history, the file transfer speed, and the like. By setting the determined data range in a rule that is referred to when the open mapping information return range determination unit 33 determines the mapping information range, acquisition of invalid mapping information can be avoided.

  FIG. 12 shows a file management system according to the second embodiment of the present invention. The configuration of this embodiment is a configuration in which a mapping information acquisition logging unit 37 and a rule determination unit 38 are added in the meta server 30 to the configuration of the first embodiment (FIG. 1). The mapping information acquisition logging unit 37 collects a history of mapping information acquisition requests transmitted from the host 10 that arrive at the meta server 30. The rule determination unit 38 refers to the history collected by the mapping information acquisition logging unit 37 to check access to the file after the file open, and the mapping information return range determination unit 33 at the time of open returns the mapping information returned at the time of the file open. Create a rule to use when determining the range.

  The rule determination unit 38 checks the data area accessed by the host 10 from the history of the mapping information acquisition request transmitted by the host after the file open. For example, when the host 10 transmits a mapping information acquisition request five times in order from the file head side after the file open, the rule determination unit 38 sets the data range for five data blocks from the file head to the host 10 at the time of opening. Create a rule for the data range of the mapping information to be returned. The open mapping information return range determination unit 33 refers to the created rule and determines the range of mapping information to be returned to the host 10 when the file is open. By doing so, mapping information of a data area that is highly likely to be accessed by the host 10 after the file open can be transmitted to the host 10 at the time of the file open. Other effects are the same as those of the first embodiment.

  Although the invention has been particularly shown and described with reference to illustrative embodiments, the invention is not limited to these embodiments and variations thereof. It will be apparent to those skilled in the art that various modifications can be made to the present invention without departing from the spirit and scope of the invention as defined in the appended claims.

This application is based on and claims the priority of Japanese Patent Application No. 2008-055095 filed on Mar. 5, 2008, the entire disclosure of which is incorporated herein by reference. join.

Claims (16)

A file management method used in a file system having a meta server for managing mapping information for managing allocation of data constituting a file to a storage device,
When the metaserver receives a file open request from the host, obtaining mapping information of the file to be opened from a mapping information storage unit that stores mapping information;
The metaserver sending the acquired mapping information to the host;
A file management method comprising: the host registering mapping information transmitted from the meta server when the file is opened in a mapping information cache.   2. The file management method according to claim 1, wherein in the step of transmitting the mapping information to the host, the meta server includes the acquired mapping information in a response to the file open request and transmits the mapping information to the host. Obtaining the mapping information comprises:
When the metaserver receives a file open request from the host, determining a range of mapping information to be sent to the host;
The file management method according to claim 1, wherein the meta server includes a step of acquiring mapping information in the determined range from the mapping information storage unit.   The meta server, in the step of determining a range of the mapping information, determines mapping information corresponding to a data range that the host is likely to access after opening a file as a range of mapping information to be transmitted to the host. Item 5. The file management method according to Item 4.   The mapping information includes a data range in the file and an allocation to the storage device, and in the step of determining the mapping information range, the metaserver sends a data range in the file to be opened to the host. In the step of determining the range of mapping information to be transmitted and obtaining the mapping information, the mapping information including the determined data range in the file in the mapping information file from the mapping information storage unit The file management method according to claim 4, wherein the file management method is acquired.   In the step of determining the range of the mapping information, the metaserver includes at least one of a data range for a predetermined size from the beginning of the file and a data range for the predetermined size from the end of the mapping information to be transmitted to the host. The file management method according to claim 6, wherein the file management method is determined as a range.   The meta server determines a data range in the file with reference to a rule that defines a data range to be transmitted to the host when a file is opened in the step of determining the range of the mapping information. File management method. The meta server records a history of mapping information acquisition requests sent by the host during file access; and
The file management method according to claim 8, further comprising: a step of determining a data range to be transmitted to the host when the file is opened in the rule by referring to the history. A meta server that manages mapping information for managing allocation of data constituting a file to a storage device,
When there is a file open request from the host, a mapping information acquisition unit that acquires mapping information of the file to be opened from a mapping information storage unit that stores the mapping information;
A mapping information transmitting unit that transmits the mapping information acquired by the mapping information acquiring unit to the host;
An open time mapping information return range determination unit that determines a range of mapping information to be transmitted to the host when the file is opened;
The mapping information acquisition unit is a meta server that acquires mapping information in a range determined by the open mapping information return range determination unit from the mapping information storage unit.   The meta server according to claim 10, wherein the mapping information transmission unit transmits the response to the file open request including the mapping information acquired by the mapping information acquisition unit to the host.   The mapping information return range determination unit at the time of opening determines mapping information corresponding to a data range that the host is likely to access after opening a file as a range of mapping information to be transmitted to the host. Meta server.   The mapping information includes a data range in the file and an allocation to the storage device, and the open time mapping information return range determination unit maps the data range in the file to be opened to the host to be transmitted. The mapping information acquisition unit determines the data range in the file determined by the open mapping information return range determination unit from the mapping information storage unit as the data range in the mapping information file. The meta server according to claim 12 or 13, wherein the mapping information is acquired.   The open mapping information return range determination unit determines at least one of a data range of a predetermined size from the beginning of the file and a data range of a predetermined size from the end as a mapping information range to be transmitted to the host. The metaserver according to claim 14.   The meta server according to claim 14 or 15, wherein the open mapping information return range determination unit determines a data range in the file with reference to a rule that defines a data range to be transmitted to the host when a file is opened. . A history generation unit for recording a history of a mapping information acquisition request transmitted by the host during file access;
The meta server according to claim 16, further comprising: a rule generation unit that determines a data range to be transmitted to the host when the file is opened in the rule with reference to the history. A meta server for managing mapping information for managing allocation of data constituting a file to a storage device, and when there is a file open request from a host, mapping information of the file to be opened from the mapping information storage unit for storing the mapping information A meta information server having a mapping information acquisition unit that acquires the mapping information transmission unit that transmits the mapping information acquired by the mapping information acquisition unit to the host;
A file management system comprising: a host having a mapping information cache that registers mapping information transmitted from the metaserver in the cache when the file is opened.

Images