This application relates to and claims the benefit of priority from Japanese Patent Application number 2010-94059, filed on Apr. 15, 2010 the entire disclosure of which is incorporated herein by reference.
The present invention relates to the allocation of a storage area to a virtual logical volume that conforms to Thin Provisioning.
Storage tier management technology that combines Thin Provisioning technology is known. In this storage tier management technology, the medium from which a real area (a substantive storage area) is allocated is selected using a virtual logical volume (hereinafter, a virtual volume) unit that conforms to Thin Provisioning. For example, according to Japanese Patent Application Laid-open No. 2007-66259, one or more media are associated beforehand with each virtual volume, and when there is a write to the virtual volume, a real area is allocated from any of the one or more media associated with this virtual volume.
There are cases where a single virtual volume is used with a plurality of either applications or host computers, and, in addition, where the performance requirements with respect to the virtual volume will differ for each either application or host computer. For instance, two specific examples of this may be considered.
As a first specific example, a case in which a single virtual volume is used by an online processing application, which must process numerous requests at high speed, a tabulation application, which batch processes the data of this application at night, and a backup application, which regularly creates backups, will be considered. In this case, it is preferable that a real area inside a high-speed medium should be allocated to a virtual area (a virtual storage area inside the virtual volume) used by the online processing application, and real areas inside media that are lower speed than this high-speed medium should be allocated to the virtual areas that are used by the other applications.
However, according to the above-mentioned technology, since the medium that serves as the source of the initially allocated real area is determined in virtual volume units, it is not always possible to allocate an appropriate real area to the virtual volume.
In the first specific example, when the setting is such that initially a real area is always allocated from a high-speed medium, a real area from a high-speed medium is also allocated to a virtual area that is the write destination of an application other than the online processing application. For this reason, over-spec (that is, the allocation of media that are too fast) occurs. By contrast, when the setting is such that initially a real area is always allocated from a low-speed medium, a real area from a low-speed medium is also allocated to a virtual area that is the write destination of the online processing application. For this reason, a performance deficiency occurs with respect to an access from the online processing application.
As a second specific example, host virtualization technology (server virtualization technology) will be considered. In host virtualization technology, a plurality of virtual volumes, which are provided from a storage apparatus, are regarded as a single virtual storage pool called a “data store”, and a virtual disk based on the data store is provided to a virtual host. This means that a many-to-many relationship is able to exist between the storage apparatus-provided virtual volume and the virtual disk that is provided to the virtual host, in other words, the storage apparatus-provided virtual volume may be used by a plurality of virtual hosts having different applications.
Accordingly, an object of the present invention is to enable an appropriate medium, which is the source of a real area that is initially allocated to a virtual volume used by a plurality of applications, to be selected for each application, virtual host or host.
The virtual volume is a virtual logical volume that conforms to Thin Provisioning, and is a logical volume configured from a plurality of virtual areas and used by a plurality of applications. In a case where the storage apparatus receives a write request comprising write-destination information for identifying a write-destination virtual area in the virtual volume, and, in addition, the write-destination virtual area is an unallocated virtual area, the storage apparatus selects a medium, which corresponds to the write to the write-destination virtual area and/or the identification information of the source of this write, from a plurality of media, which have different performances and which are each configured from two or more real areas, and allocates a real area from the selected medium to the write-destination virtual area.
An embodiment of the present invention will be explained below. In the following explanations, various types of information may be explained using the expression “xxx table”, but these various types of information may be expressed using a data structure other than a table (for example, a queue). To show that the information is not dependent on the data structure, “xxx table” may be called “xxx information”.
Further, an ID, a WWN (World Wide Name), and a name will be used as information for identifying various targets in the following explanation, but these are interchangeable and other types of identification information (for example, a number) may be used.
In the following explanation, there may be instances when “program” is used as the subject in explaining a process, but since a prescribed process is performed in accordance with a program being executed by a processor (for example, a CPU (Central Processing Unit)) while using a storage resource (for example, a memory) and/or a communication interface device (for example, a communication port) as needed, the subject of the process may also be the processor. A process that is explained having the program as the subject may be a process that is performed by a management system. The processor may comprise a hardware circuit for processing either all or a portion of the processes performed by the processor. A computer program may be installed in respective computers from a program source. The program source, for example, may be a program delivery server or a storage medium.
A management system may be configured using one or more computers. Specifically, for example, in a case where a management computer displays information, or a case where the management computer sends display information to a remote computer, the management computer is the management system. Further, for example, in a case where the same functions as those of the management computer are realized by a plurality of computers, this plurality of computers (may include a display computer in a case where the display computer carries out displays) is the management system.
Further, date information may be included in a “time” stated in the present invention. In addition, the format for expressing “time” data may be expressed as the number of seconds, milliseconds, and microseconds from a predetermined starting point, may be expressed in the format year, month, day, hour, minute and second, or may be expressed using another format. Further, the “time” may refer to a specific point-in-time, and may possess duration as in a time period.
In the present invention, “application” refers to a series of either virtual host- or host-performed operations, which are stipulated in accordance with the following (A) and (B):
(A) one or more application programs; and
(B) at least one of (b1) setting information of this program, (b2) data targeted for processing by this program, (b3) content of the processing request received by his program, and (b4) frequency with which this program receives a processing request. One example thereof is the above-mentioned online processing, tabulation processing or backup processing. Furthermore,
A host computer (hereinafter, simply called the host) 101 and a management system 103 are coupled to a storage apparatus 109.
The management system 103 manages the storage apparatus 109. The management system 103 may also comprise either the storage apparatus 109 or the host 101.
The storage apparatus 109 comprises a pool 1613 and a plurality of virtual volumes (hereinafter, VVOL) 1615.
The pool 1613 is configured from a plurality of tiers. In other words, the pool 1613 is configured from a plurality of media 1611 having different performances. The tier height depends on the performances of the media 1611. For example, a first medium (the fastest medium) 1611A, which has the highest performance among the plurality of media 1611, is equivalent to the highest tier, and a third medium (the slowest medium) 1611C, which has the lowest performance among the plurality of media 1611, is equivalent to the lowest tier. As used here, “performance”, for example, is access performance. Access performance, for example, includes response time (for example, the time it takes a medium-based physical storage device to output a response to a controller after receiving a data access request from the controller) and a data transfer rate (the amount of data inputted/outputted per unit of time).
Each medium 1611 is configured from two or more real areas 1614. Therefore, the pool 1613 is configured from a plurality of real areas (substantive storage areas) 1614. Each medium 1611 is a cluster of one or more real volumes based on one or more physical storage devices (for example, a RAID (Redundant Array of Independent (or Inexpensive) Disks) group, which is configured from a plurality of physical storage devices). A “real volume” is a logical volume based on a physical storage device. The performance of a medium 1611 depends on at least one of the type of physical storage device that forms the basis for this medium 1611, a RAID level or a RAID combination (a configuration like 4D+2P or 3D+1P). As types of physical storage devices, for example, there are a SSD (Solid State Drive), a SAS (Serial Attached SCSI)—HDD (Hard Disk Drive), and SATA (Serial ATA)—HDD. Furthermore, SSD differs from a HDD in that it is a physical storage device that uses a semiconductor memory, such as a flash memory or DRAM.
The VVOL 1615 is configured from a plurality of virtual areas (virtual storage areas) 1616.
Host virtualization technology (server virtualization technology) is applied to the host 101. The host 101 comprises a plurality of virtual hosts 213 and a data store 1621. The data store 1621 is a cluster of a plurality of mounted volumes 1625. The mounted volume 1625 is a storage space that corresponds to a VVOL 1615 that is mounted from the storage device 109. The mounted volume 1625 and the VVOL 1615 correspond to each other on a one-to-one basis. One part of the data store 1621 is provided to a virtual host 213 as a logical volume 1623. In accordance with a configuration like this, a many-to-many relationship is able to exist between the VVOL 1615 and the logical volume 1623, in other words, the VVOL 1615 is able to be used by a plurality of virtual hosts 213 having different applications.
Furthermore, “mounted” indicates at the least that the host 101 recognizes the VVOL 1615 and creates management information required for creating the data store 1621. In addition, as an example of the above-mentioned management information, there is the correspondence between the VVOL 1615 identifier and the identifier allocated for the host 101 to manage the VVOL 1615. Examples of VVOL 1615 identifiers may include (1) the storage apparatus identifier and a unique identifier inside the storage apparatus that is allocated to the VVOL 1615, and (2) a port identifier (for example, at least one of a WWN, a port ID, and an IP address) of the storage apparatus and a LUN (Logical Unit Number) allocated to the VVOL 1615. In the following explanation, the host 101 will be explained using the term “VVOL 1615”, but this explanation is about the VVOL 1615, that is, the mounted volume 1625 that is recognized by the host 101.
The allocation of a real area to the virtual area 1616 includes an “initial allocation” and a “allocation in reallocation”.
The “initial allocation” is to allocate an unallocated real area (a real area that is not allocated to any virtual area 1616) 1614 to an unallocated virtual area 1616. Therefore, the “initial allocation” may be carried out upon receiving a write request that specifies an address range (for example, a LBA (Logical Block Address) and a write data length), which comprises this unallocated virtual area inside the VVOL. Further, the “initial allocation”, for example, may also be carried out when a migration of data from a VVOL 1615 to another VVOL 1615 triggers the need to store data in this unallocated virtual area. Furthermore, in the above explanation, the real area to be allocated is assumed to be an unallocated area at the time of allocation to the virtual area, but this does not always have to be so. That is, an area that has already been allocated to another virtual area may be allocated to the unallocated virtual area at allocation time.
An “unallocated virtual area” is either a virtual area to which a real area is not allocated, or a virtual area to which a real area is not substantially allocated (for example, a virtual area to, which a special real area with prescribed data (for example, a real area in which the value of all the bits is “0”) is allocated). In a case where a write occurs to an unallocated virtual area, and a special real area is allocated to this virtual area, an unallocated real area (a real area that is not allocated to a virtual area) is allocated to this virtual area in place of the special real area, and the write data is written to this real area. There may be n (where n is a natural number) special real areas in a single pool. Further, a special real area may be shared by a plurality of unallocated virtual areas.
The “allocation in reallocation” is when, in a reallocation process, a data migration-destination real area 1614 is allocated in place of a migration-source real area 1614 to a virtual area 1616 to which the data migration-source real area 1614 has been allocated. In this embodiment, an appropriate medium 1611 is selected as the source of the real area that is allocated at the initial allocation as described above. Furthermore, in the case of a “allocation in reallocation”, it is typical for data stored in the migration-source real area 1614 to be moved (or copied) to the migration-destination real area.
In a write from a virtual host 213, for example, the following (a1) through (a7) are carried out.
(a1) A program, which is virtually executed by the virtual host 213, requests a write to a virtual disk (called the virtual host disk) 1623 that is provided to this virtual host 213. The write request at this time specifies a range of addresses (the virtual host disk address range) on the virtual host disk 1623. Furthermore, in the case of a write request, the virtual host 213 also creates write-targeted data (write data). The virtual host 213 is a virtual computer provided in accordance with the host 101 processing a hypervisor program, and similarly the operating host disk is a virtual disk that is provided to the program, which is being virtually executed on the virtual host, in accordance with processing the hypervisor program.
(a2) The hypervisor program of the host 101 sends the write request and write data that specify the address range of the VVOL 1615 corresponding to the virtual host disk address range (for example, the LBA (Logical Block Address) and the write data length) to the storage apparatus 109 in response to the request of (a1).
(a3) The storage apparatus 109 receives the write request sent in (a2). The storage apparatus 109 comprises a cache memory area, and stores the write data that conforms to the write request in the cache memory area. When the write data is written to the cache memory area, the storage apparatus 109 may respond to the host 101 with a write-complete report.
(a4) The storage apparatus 109 identifies from the write request-specified address range received in (a3) a virtual area (hereinafter, the write-destination virtual area) 1616 included in this address range.
(a5) The storage apparatus 109 determines whether or not the write-destination virtual area 1616 is an unallocated virtual area.
(a6) In a case where the write-destination virtual area 1616 is an unallocated virtual area, the storage apparatus 109 carries out an initial allocation process. In the initial allocation process, the storage apparatus 109 identifies a medium 1611 that corresponds to the write-source application, the host 101, or the virtual host 213, and allocates an unallocated real area 1614 from the identified medium 1611 to the write-destination virtual area 1616.
(a7) The storage apparatus 109 writes the write-targeted data to the real area 1614 that has been allocated to the write-destination virtual area 1616. The storage apparatus 109 may respond to the host 101 with a write-complete report when the write-targeted data has been written to the real area 1614 instead of when the write-targeted data has been written to the cache memory area.
On the other hand, in a read from a virtual host 213, for example, the following (A1) through (A5) are carried out.
(A1) A program, which is virtually executed by the virtual host 213, requests a read with respect to the virtual host disk 1623 that is provided to this virtual host 213. The read request at this time specifies the virtual host disk address range.
(A2) The host 101 hypervisor program sends the read request specifying the address range of the VVOL 1615 corresponding to the virtual host disk address range to the storage apparatus 109 in response to the request of (A1).
(A3) The storage apparatus 109 receives the read request sent in (A2). The storage apparatus 109 specifies the address range specified in the read request, reads data that was written in the past (called the read-targeted data or the read data) from either the cache memory area or an allocated real area, and sends this data to the host 101.
(A4) The host 101 hypervisor program receives the above-mentioned read data, identifies the virtual host that requested the read in (A1), and virtually sends the read data to this virtual host.
(A5) A program, which is virtually executed by the virtual host, receives the read data of (A4) as read data that corresponds to the requested read.
Examples of the “program, which is virtually executed by the virtual host” in the above explanation are an OS and an application program. Furthermore, the above explanation that has the “storage apparatus” as the subject may be considered a process, which is carried out by a controller that is included in the storage apparatus.
Furthermore, the process in which the address range in the virtual volume 1616 is identified from the virtual disk address range, which is the target of the write and read of the above-mentioned (a2) and (A2), is carried out based on conversion information of the host 101.
The initial allocation process related to this embodiment is based on the concept that the write-source application, virtual host, or host correspond to write-source identification information and/or a write time period.
The write-source identification information is for identifying the write source, and in this embodiment, is the WWN (World Wide Name) of the write-source virtual host 213. In a case where host virtualization technology is not applied (a case in which there are no virtual hosts 213 in the host 101), the write-source identification information may be the host 101 WWN (specifically, the WWN allocated to the interface device (for example, the HBA (Host Bus Adapter)) of the host 101).
The write time period, for example, is the time period to which the write time belongs. The write time, for example, is the time at which the write request was received, the time at which the real area was allocated in response to the write request, or the time displayed by the timestamp of the write request.
As the method for identifying the write-source application, for example, a method in which either the storage apparatus 109 or the management system 103 has an agent program, and the agent program queries the application at the time of the write from the host 101 is conceivable. However, this method is not realistic.
Accordingly, the write-source identification information and/or the write time are used in this embodiment in order to identify the write-source application, virtual host, or host. This is based on at least one of the following plurality of assumptions.
In the case of this method, the storage apparatus 109 is substantially able to identify the write-source application on its own.
Furthermore, in the present invention, the write-source identification information, for example, is either the identifier of the WWN (or a port number created based on the WWN) of the interface device of the write-request source host 101 (or virtual host 213) included in the write request, or information obtained by converting this information.
In this embodiment, initial allocation policy information is prepared beforehand. Based on the initial allocation policy information, the medium 1611 corresponding to the write-source identification information and/or the write time is identified in the above-mentioned (a6).
In the initial allocation policy information, first a medium (hereinafter, the initial medium) 1611, which will become the source of the real area, is defined in VVOL 1615 units. Then, the initial allocation policy information is updated in a timely manner based on the result of the reallocation process. In the reallocation process, a data migration is performed between real areas 1614. The real area 1614, which has been allocated to the virtual area 1616, becomes the migration-source real area 1614, and a real area 1614 inside a medium 1611 that differs from the medium 1611 comprising this real area 1614 becomes the migration-destination real area 1614. In the reallocation process, whether or not the real area 1614 that has been allocated to the virtual area 1616 becomes the target of the data migration is determined on the basis of the access status of this virtual area 1616. The above-mentioned “access” includes the above-described read and/or write. The access status, for example, may be the last access time (the last write time and/or the last read time) or the access load. The access load, for example, may be either the data transfer rate (the amount of data inputted/outputted per unit of time), or the access frequency. In this embodiment, whether or not the real area 1614 that has been allocated to the virtual area 1616 becomes the target of the data migration is determined on the basis of the access frequency of this virtual area 1616.
Specifically, in this embodiment, for example, the following (Item 1) and (Item 2) are for the reallocation process.
(Item 1) An access frequency range is defined for either each medium 1611 or each tier.
(Item 2) The storage apparatus 109 manages the access frequency (for example, the unit “IOPS”) for each virtual area 1616. For example, when a write and/or read is carried out with respect to the virtual area 1616 (or, when a write and/or read is carried out with respect to the real area that is allocated to the virtual area 1616), the storage apparatus 109 updates the access frequency corresponding to this virtual area 1616. The storage apparatus 109 either regularly or repeatedly sets the access frequency for each virtual area 1616 to zero, for example.
In the above items, the access frequency is used as an example, but a range of access statuses assumed for either each medium 1611 or each tier may be defined using another access status, and the storage apparatus 109 may manage the access status for each virtual area 1616.
The reallocation process is carried out either regularly or irregularly. In the reallocation process, for example, the following (b1) and (b2) are carried out.
(b1) The storage apparatus 109 determines whether or not the access frequency of the virtual area 1616 exceeds the access frequency range corresponding to the medium 1611 comprising the real area 1614 that has been allocated to this virtual area 1616 for each virtual area 1616. In the case of another access status, this process determines whether or not the access status of each virtual area 1616 falls outside the range of the assumed access status range corresponding to the medium 1611 comprising the real area 1614 that has been allocated to this virtual area 1616.
(b2) For a virtual area (hereinafter, called the “target virtual area” in the explanation that refers to
(b21) The storage apparatus 109 migrates data from a migration-source real area to an arbitrary unallocated real area (a migration-destination real area) inside a different medium 1611 than the medium 1611 that comprises this migration-source real area. The “migration-source real area” is the real area that has been allocated to the target virtual area 1616. The medium 1611 that comprises the migration-destination real area is the medium 1611 corresponding to the access frequency range into which the frequency range of the target virtual area 1616 falls. In the case of another access status, the medium 1611 that comprises the migration-destination real area will be the medium 1611 corresponding to the assumed access status range within which the access status of the target virtual area 1616 falls.
(b22) The storage apparatus 109 allocates the migration-destination real area to the target virtual area 1616 in place of the migration-source real area.
The management system 103, based on the result of the reallocation process, updates the initial allocation policy information. For example, in the following case, the management system 103 updates the initial allocation policy information such that the medium, which corresponds to a first write-source identification information and/or a first write time period, transitions from the first medium (for example, a SSD medium) 1611A to the second medium (for example, a SAS medium) 1611B.
The host 101 and the storage apparatus 109 are coupled to a SAN (Storage Area Network) 105. The storage apparatus 109 and the management system 103 are coupled to a LAN (Local Area Network) 107. At least one of the storage area network 105 and the network 107 for management use may be a different type of communication network. The storage area network 105 and the network 107 for management use may be the same type of network, and may be integrated together.
The host 101 sends an access request to the storage apparatus 109. The storage apparatus 109 receives the access request from the host 101 via the SAN 105, and processes this access request. An example of the access request is the above-mentioned read request and/or write request.
The management system 103 sends a control request (for example, an initial allocation policy information update request) to the storage apparatus 109. The storage apparatus 109 receives the control request from the management system 103 via the LAN 107, and carries out processing in accordance with the control request.
Furthermore, the computer system may comprise a plurality of hosts 101. Similarly, the computer system may comprise a plurality of storage apparatuses 109.
The host 101 is configured from one or more computers. The host 101 comprises a SAN port 203, a memory 205, and a CPU (Central Processing Unit) 201, which is coupled to the memory 205 and the SAN port 203. Another type of storage resource may be used either instead of or in addition to the memory 205.
The SAN port 203 is coupled to the SAN 105, and, for example, a WWN, which is the identification information of the SAN port 203, is allocated to the SAN port 203 of the HBA (Host Bus Adapter) built into the host 101.
The memory 205 stores the above-mentioned hypervisor program 211 (may be abbreviated as hypervisor), and one or more virtual host information 213i. The hypervisor 211 is a computer program for host virtualization. One or more virtual hosts (virtual computers) 213 are run on the host 101 in accordance with the hypervisor 211 being executed by the CPU 201 based on the above-mentioned virtual host information.
Each virtual host 213 carries out communications using the SAN port 203. A virtual WWN, which is the identification information of the virtual port, is allocated to each virtual host 213. An access request from the virtual host 213, for example, comprises the virtual WWN and access-destination information. The access destination information is for identifying the access-destination area, and, for example, comprises a LUN (Logical Unit Number) and a LBA (Logical Block Address). It is possible to identify the access-destination VVOL from the LUN and to identify the virtual area inside this VVOL from the LBA. The access request from the virtual host 213 is outputted via the SAN port 203, and reaches the storage apparatus 109 via the SAN 105.
The virtual host information 213i is information that stores the setting content of the virtual host, and an example of the setting content is the above-mentioned virtual WWN and/or the capacity of the virtual host disk.
The access-destination information may also comprise information (for example, a WWN and/or a port number) for identifying either the storage apparatus or the port thereof.
The storage apparatus 109 comprises a plurality of RAID groups 305 of different performance, and a controller, which is coupled to the plurality of RAID groups 305, and which controls access to a RAID group 305.
The RAID group 305 is configured from a plurality of the same type of physical storage devices. For example, as the RAID group 305 there may be a RAID group configured from SSD, a RAID group configured from SAS-HDD, and a RAID group configured from SATA-HDD. Various physical storage devices may be used as the physical storage device, such as a HDD and a flash memory device. Furthermore, the above-mentioned coupling of the controller and the RAID group 305 actually refers to the coupling of the controller and the physical storage device.
A logical volume, that is, a real volume (hereinafter, RVOL) 306 which is based on the RAID group 305, is provided. One medium is configured from one or more RVOL 306 that are based on the RAID group 305 of the same performance. The performance of the medium depends on the performance of the RAID group 305 on which the RVOL 306 configuring this medium is based. The performance of the RAID group 305, for example, depends on at least one of the type of physical storage device (the type of physical storage device that configures the RAID group 305), the RAID level and the RAID combination (a configuration like 4D+2P or 3D+1P).
The medium is a component of a pool. For this reason, the medium is configured from two or more real areas. Therefore, the pool is configured from a large number of real areas. One or a plurality of pools may exist in a single storage apparatus. Further, there may be a real volume that does not belong to any pool.
The controller comprises a SAN port 302, a LAN port 303, a memory 304, and a CPU 301 that is coupled to these elements. Another type of storage resource may be used either instead of or in addition to the memory 304. The memory 304 also comprises a cache memory area. Either read data or write data for the RAID group 305 is temporarily stored in the cache memory area.
The SAN port 302 is coupled to the SAN 105. The SAN port 302 receives an access request from the virtual host 213.
The LAN port 303 is coupled to the LAN 107. The LAN port 303 receives a control request from the management system 103.
The memory 304 stores a storage control program 311, a VVOL management table 312, a real area management table 313, a RAID group management table 314, a storage tier definition table 315, and an initial allocation policy table 316. The storage control program 311 is able to carry out an access control process and a reallocation process in accordance with being executed by the CPU 301.
The storage apparatus in the present invention may be of any packaging mode as long as it includes a controller and a physical storage device. For example, the storage apparatus may be a single apparatus as shown in
Next, the tables 312 through 316 will be explained.
The VVOL management table 312 denotes which real area is allocated to which virtual area of the VVOL, denotes the access-source WWN of the write that triggers the allocation of a real area to the virtual area in the initial allocation process and when it occurs, and denotes the medium from inside which the real area has been allocated to the virtual area in the reallocation process. Specifically, for example, the VVOL management table 312 comprises the following information for each virtual area.
According to table 312 of
In the example of
The real area management table 313 denotes whether or not a real area is unallocated. Specifically, for example, the real area management table 313 comprises the following information for each real area.
The RAID group management table 314 comprises information related to the performance of the RAID group. Specifically, for example, the RAID group management table 314 comprises the following information for each RAID group.
Another type of information (for example, information denoting the RAID group combination) may be included either instead of or in addition to at least one of the pieces of information 701 through 703 mentioned above. Furthermore, the information in the table shown in the drawing is an example, and this fact is not limited to
The storage tier definition table 315 comprises information related to the performance of the medium (tier) and the access frequency range. Specifically, for example, the storage tier definition table 315 comprises the following information for each medium (tier).
The initial allocation policy table 316 is an example of the above-mentioned initial allocation policy information. The initial allocation policy table 316 denotes either the medium or the tier corresponding to the access-source WWN and the access time period. Specifically, for example, the initial allocation policy table 316 comprises the following information for each access-source WWN and access time period combination.
According to the table 316, a combination of an access-source WWN and a time period corresponds to an application. Therefore, determining an initial medium (tier) for each access-source WWN and time period combination determines the initial medium (tier) for each application.
Furthermore, it is possible to set an asterisk value “*”, which indicates all cases, for both the access-source WWN 902 and the time period 903. In other words, either the access-source WWN 902 or the time period 903 alone may decide the initial medium. Also, in
Furthermore, in a case where
Furthermore, because the initial allocation policy table 316 is referenced in a case where an unallocated virtual area is detected in accordance with a write request, it is preferable that the table size be made as small as possible in order to rapidly determine the initial allocation tier. As a measure for achieving this, a combination of information 901, 902, and 904 for which the current time is included in the time period 903 may be obtained from
The management system 103 is configured from one or more computers. The management system 103 comprises a LAN port 403, a memory 402, a display device (for example, a liquid crystal display) 451, and a CPU 401 that is coupled to these. Another type of storage resource may be used either instead of or in addition to the memory 402.
The memory 402 stores a storage management program 411, a host management table 317, an application definition table 318, a reallocation log table 412, and an initial allocation policy table 413. The following processing is carried out in accordance with the storage management program 411 being executed by the CPU 401.
The configuration of the initial allocation policy table 413 is the same as the configuration of the initial allocation policy table 316 of the storage apparatus 109. The storage management program 411 updates the initial allocation policy table 316 such that the information of the initial allocation policy table 316 is identical to the information of the post-update initial allocation policy table 413.
The host management table 317 denotes the relationship between the virtual host and the WWN. Specifically, for example, the host management table 317 comprises the following information for each virtual host.
Furthermore, the hostname 1001 and the WWN 1002 can be acquired from a program being executed on the virtual host or from the host hypervisor program, but another acquisition method, such as manual input, may also be used.
The application definition table 318 denotes the definition of the application. Specifically, for example, the application definition table 318 comprises the following information for each application.
The application name, the virtual host on which the application is to be executed, and the time period during which the application is to be executed can be acquired from either the program being executed on the virtual host or from the host hypervisor program, but another acquisition method, such as a manual input, may also be used.
Further, an asterisk value “*”, which indicates all cases, may be used in one of the hostname 1102 or the time period 1103 for each application. This signifies that the application definition does not depend on either the virtual host on which the application is to be executed or the time period during which the application is to be executed.
Furthermore, in a case where
The reallocation log table 412 comprises information denoting a log of the results of reallocation processing. Specifically, for example, the reallocation log table 412 comprises the following information for each combination of an access-source WWN and a time period.
According to the table 412 of
Taking into account the rate of occurrence of each of the cases mentioned above, it is clear that the medium “Tier 3” is appropriate as the initial medium corresponding to the virtual host “WWN-C” and the time period “12:00-13:00 on Dec. 25, 2009”. Therefore, in a case where the value of the initial allocation tier 1205 corresponding to the virtual host “WWN-C” and the time period “12:00-13:00 on Dec. 25, 2009” is something other than “Tier 3”, it is preferable that this value to updated to “Tier 3”. This is because the amount of data migrated in the reallocation process should be as small as possible.
The processes carried out by this embodiment, for example, are (1) the initial setup, (2) the write process, (3) the reallocation process, and (4) the updating of the initial allocation policy table. Each of these processes will be explained hereinbelow.
The storage management program 411 inside the management system 103 displays an initial setup screen 1501, an example of which is given in
Prior to a VVOL being provided to the host 101 from the storage apparatus 109 (prior to a VVOL being mounted to the host 101), the storage administrator uses the initial setup screen 1501 to select the initial medium (tier) corresponding to this VVOL. That is, at least the VVOL ID and the initial medium (tier) are selected. The storage management program 411 registers the selected VVOL ID and initial medium (tier) ID in the initial allocation policy table 413 inside the management system 103. The storage management program 411 also has the storage control program 311 inside the storage apparatus 109 register this VVOL ID and medium ID in the initial allocation policy table 316 inside the storage apparatus 109.
In the initial setup, in a case where the access-source WWN and the hostname are selected, and, in addition, the application name and the time period are inputted via the initial setup screen 1501, the storage management program 411 is able to register the access-source WWN and the time period in the initial allocation policy table 413. Further, the storage management program 411 is able to have the storage control program 311 register the access-source WWN and the time period in the initial allocation policy table 316. The storage management program 411 is also able to have the storage control program 311 register a combination of the application name, the hostname, and the time period in the application definition table 318.
The storage administrator does not have to input at least one of the access-source WWN and the time period. In accordance with this, the prescribed code “*” is registered as the value of the access-source WWN 902 and/or the time period 903 in the initial allocation policy table 316. The “*” here signifies that the conditions will be satisfied no matter what the access-source WWN and/or time period (the real area will be allocated from the defined initial medium).
In 51001, the virtual host issues a write request. This write request comprises the virtual host WWN and write-destination information. The write-destination information comprises a LUN and the virtual area LBA. The write-destination information may also comprise the data size of the data (the write-targeted data) attached to the write request.
In S1002, the storage apparatus 109 receives the write request from the virtual host. The program 311 writes the write-targeted data to the cache memory area. Further, the storage control program 311 identifies the VVOL ID corresponding to the LUN of the write-destination information in the write request, and, in addition, identifies the real area ID 503 corresponding to the write-destination information from the VVOL management table 312.
In a case where the identified real area ID 503 is a valid value (a value denoting the ID of any real area) (S1003: YES), the program 311 writes the write-targeted data that is in the cache memory area to the real area identified from the identified real area ID 503 in S1002.
In a case where the identified real area ID 503 is an invalid value (a value signifying that the real area is not allocated) (S1003: NO), in S1004, the program 311 determines on the basis of the real area management table 313 whether or not an unallocated real area that is capable of being allocated to all the write-destination virtual areas (the virtual areas identified from the write-destination information) exist in the pool.
In a case where the result of the determination of S1004 is negative (S1004: NO), in S1005, the program 311 returns an error to the virtual host that was the source of the write request.
In a case where the result of the determination of S1004 is affirmative (S1004: YES), in S1006, the program 311 identifies the initial medium ID (the initial allocation tier 904) corresponding to the VVOL ID identified using the LUN inside the write-destination information, the virtual host WWN and the time at which the write request was received from the initial allocation policy table 316. Next, in S1007, the program 311 identifies the medium performance 802 corresponding to the identified initial medium ID from the storage tier definition table 315. Next, in S1008, the program 311 identifies the RAID group ID 701 of the RAID group that conforms to the PDEV type and RAID level denoted in the identified medium 802 from the RAID group management table 314.
In S1009, the program 311, using the ID that denotes the identified RAID group IS 701 as a key, determines from the real area management table 313 whether or not the plurality of real areas based on the RAID group identified from this ID include unallocated real areas that are capable of being allocated to all the write-destination virtual areas.
In a case where the result of the determination of S1009 is affirmative (S1009: YES), in S1010, the program 311 allocates the unallocated real area inside the initial medium to the write-destination virtual area, and in S1012, writes the data inside the cache memory area to this real area. Specifically, in S1010 the program 311 carries out the following updates.
In a case where the result of the determination of S1009 is negative (S1009: NO), in S1011, the program 311 allocates as many real areas as possible from the initial medium to the write-destination virtual area, allocates only enough real areas from another medium to the write-destination virtual area to make up the shortage, and thereafter, in S1012 writes the data inside the cache memory area to the allocated real areas.
Specifically, for example, in the above-mentioned S1011, in a case where there are 20 write-destination virtual areas and only 13 unallocated real areas in the initial medium, the program 311 allocates the 13 real areas to 13 write-destination virtual areas, and allocates 7 real areas from another medium to the remaining 7 write-destination virtual areas. As the “other medium” that serves as the source of real areas here is the medium that most closely approximates the performance of the initial medium (for example, the tier adjacent to the initial medium (tier)). In a case where there are adjacent media both above and below the initial medium at this time, the medium that has the most unallocated real areas may be selected as the “other medium” that serves as the source of the real areas. This is because selecting the adjacent medium with fewer unallocated real areas will increase the likelihood of this medium having a shortage of unallocated real areas when this medium is used as the initial medium.
Furthermore, in S1011, the VVOL management table 312 and the real area management table 313 are updated the same as in S1010.
As described hereinabove, in the write process an unallocated real area is preferentially allocated from the initial medium to one or more write-destination virtual areas. Furthermore, for example, in a case where there is a write to the VVOL “Volume 1” when the status is as shown in the initial allocation policy table 316 of
Furthermore, in a case where a read request is sent from the virtual host, for example, the following processing is carried out.
(x1) The storage control program 311 receives the read request from the virtual host.
(x2) The program 311 identifies the read-source VVOL and the read-source virtual area based on the access-destination information in the read request.
(x3) The program 311 determines whether or not the read-targeted data is saved in the cache memory area.
(x4) In a case where the result of the determination of the above-mentioned (x3) is affirmative, the program 311 sends the read-targeted data of the cache memory area to the virtual host. The value of the access frequency 507 corresponding to the read-source virtual area may or may not be updated in accordance therewith.
(x5) In a case where the result of the determination of the above-mentioned (x3) is negative, the program 311 identifies the real area that is allocated to the read-source virtual area identified in the above-mentioned (x2) based on the VVOL management table 312. The program 311 reads the data from the identified real area and writes this data to the cache memory area. Then, the program 311 sends this data that is inside the cache memory area to the virtual host. The program 311 updates the value of the access frequency 507 corresponding to the read-source virtual area.
In S1101, the storage control program 311 clears the values of all the reallocation determination results 508 in the VVOL management table 312.
S1102 is carried out for all the virtual areas registered in the VVOL management table 312. In S1102, the program 311 identifies from the storage tier definition table 315 the access frequency range 603 denoting the access frequency range comprising the access frequency denoted by the access frequency 507 of the virtual area, and records the ID of the medium corresponding to this access frequency range 803 as the value of the reallocation determination result 508 corresponding to this virtual area. As a result of this, there are times when this medium ID is the same as the medium ID denoted in the tier ID 504 corresponding to this virtual area, and times when it is different.
In S1103, the program 311 arranges the virtual area identification information (for example, a combination of the VVOL ID and the LBA range) in the memory 304 in order from the access frequency 507 with the highest value.
The program 311 carries out S1104 and later steps in order (in the order in which the virtual area identification information is arranged in the memory 304) from the virtual area having the highest access frequency. S1104 and later steps will be explained hereinbelow using a single virtual area (referred to as the “target virtual area” in the explanation of
In S1104, the program 311 determines whether or not the value of the tier ID 504 corresponding to the target virtual area matches the value of the reallocation determination process 508 corresponding to the target virtual area.
In a case where the result of the determination of S1104 is affirmative (S1104: YES), the processing with respect to the target virtual area is ended since a migration is not needed.
In a case where the result of the determination in S1104 is negative (S1104: NO), the program 311 makes a determination in S1105 based on the real area management table 313 as to whether or not there is an unallocated real area inside the medium (hereinafter referred to as the “target medium” in the explanation of
In a case where the result of the determination of S1105 is affirmative (S1105: YES), the program 311 in S1107 migrates data from the real area (the migration-source real area) allocated to the target virtual area to the unallocated real area (the migration-destination real area) inside the target edium. The program 311 also allocates the migration-destination real area to the target virtual area in place of the migration-source real area. Specifically, the program 311 updates the value of the allocation status 602 corresponding to the migration-source real area to unallocated. Further, the program 311 updates the value of the real area ID 503 corresponding to the target virtual area to the ID of the migration-destination real area, and, in addition, updates the value of the tier ID 504 corresponding to the target virtual area to the ID of the target medium.
In a case where the result of the determination of S1105 is negative (S1105: NO), the program 311 determines in S1106 whether or not there is a replaceable real area. Specifically, for example, a determination is made as to whether or not there is data among the data inside the plurality of allocated real areas in the target medium that may be migrated to the real area allocated to the target virtual area.
In a case where the result of the determination of S1106 is affirmative (S1106: YES), the program 311 in S1108 replaces the real area. Specifically, for example, the program 311 carries out the following processing. The program 311 also updates the VVOL management table 312 and the real area management table 313 in accordance with the following processing.
In a case where the result of the determination of S1106 is negative (S1106: NO), the program 311 in S1109 migrates the data from the first real area to an unallocated real area inside a medium with performance that most closely approximates that of the target medium. The program 311 also updates the VVOL management table 312 and the real area management table 313 the same as in the processing of 51107.
The preceding has been an explanation of the reallocation process in a case where the access status is assumed to be the access frequency. In the case of another access status, the processing is as follows.
The storage control program 411 inside the management system 103, for example, acquires the result of the reallocation process from the storage apparatus 109 each time the reallocation process shown in
Next, based on the reallocation log table 412, the program 411 executes a process that groups together a series of time periods for which the values of the initial allocation tiers 1205 are the same into a single initial allocation policy for each access-source WWN with regard to the respective VVOL. Specifically, for example, the program 411 prepares an intermediate table 2001 in the memory 304. The program 411 registers the value of the VVOL ID 1201 in the intermediate table 2001 as the value of the VVOL 2011, and registers the value of the WWN 1202 in the intermediate table 2001 as the value of the access-source WWN 2012. The program 411 identifies from the reallocation log table 412 the record in which the value of this WWN 2012 and the value of the initial allocation tier 1205 are both the same. In a case where two or more time periods denoted by two or more identified records are consecutive (for example, when the time periods are sequential as in 9:00-10:00, 10:00-11:00, and 11:00-12:00), the program 411 groups the two or more time periods together into one time period (for example, groups the three time periods 9:00-10:00, 10:00-11:00, and 11:00-12:00 into a single time period 9:00-12:00). The program 411 registers a value denoting the time periods, which have been grouped together into one time period, in the intermediate table 2001 as the value of the time period 2013. The program 411 registers the value of the initial allocation tier 1205 as the value of the initial allocation tier 2014.
Next, the program 411 compares the values of the initial allocation tiers 2014 with the access-source WWNs that have the same time period 2013 for each VVOL. In a case where the values of the initial allocation tiers 2014 are identical to one another, the program 411 registers the plurality of WWNs as the values of the access-source WWN 902 in one record (row) of the initial application policy table 413. The VVOL ID, the time period (the time periods that have been grouped together into a single time period), and the initial medium ID are registered in this record.
The program 411 sends an update request comprising update information, which is information denoted by the post-update initial allocation policy table 413 (or information denoting the difference between the post-update initial allocation policy table 413 and the pre-update initial allocation policy table 413), to the storage apparatus 109. The storage apparatus 109 receives the update request, the storage control program 311 responds to this update request, and based on the update information of this update request updates the initial allocation policy table 316 so that the contents are the same as those of the post-update initial allocation policy table 413.
As described hereinabove, in the initial allocation policy creation process, information that is the result of grouping together common information in the reallocation log table 412 is registered in the initial allocation policy table 413, and the initial allocation policy table 316 inside the storage apparatus 109 is updated to the same contents as those of this initial allocation policy table 413. This makes it possible to keep the number of records of the initial allocation policy tables 413 and 316 to the minimum. This contributes to preventing a drop in write process performance. This is because a smaller number of records lessens the load when searching the initial allocation policy table 316. The smaller number of records also reduces the amount of capacity consumed in the memories 304 and 402.
Furthermore, a procedure other than the procedures described hereinabove (for example, data clustering) may be used as the tabulation procedure in the initial allocation policy creation process.
Further, the updating of the initial allocation policy tables 413 and 316 based on the tabulation of the results of reallocation processing may either be carried out without receiving a clear instruction from the user, or may be carried out after responding to the receipt of a clear instruction from the user. In a case where the updating of the initial allocation policy tables 413 and 316 is carried out upon receiving a clear instruction from the user, for example, an initial allocation policy definition screen 1401, which is shown in
In the screen 1401, there is displayed the pre-update initial allocation policy table 413, a table 1701 that merges the host management table 317 and the application definition table 318, a tentative post-update initial allocation policy table 413, and a table 1703 that merges the host management table 317 and the application definition table 318.
The “tentative post-update initial allocation policy table 413” is the as-yet-to-be-updated initial allocation policy table 413 showing how it is assumed the initial allocation policy table 413 will look once it is updated. The information denoting the tentative post-update initial allocation policy table 413 is displays as recommended information.
The values of the “hostname” in the tables 1701 and 1703 are the values of the hostnames 1001 corresponding to the WWNs in the table 413. The values of the “application name” in the tables 1701 and 1703 are the values of the application names 1101 corresponding to the value of the above-mentioned “hostname” and time period 903.
Information denoting one initial allocation policy is displayed in one record of table 1703, and, in addition, a check box is displayed. When a check mark is placed in the check box and the “apply” button 1705 is pressed, the program 411 registers, in the initial allocation policy tables 413 and 316, the information (the initial allocation policy) denoted in the check marked record. That is, the storage administrator is able to select whether or not the definition is to be made in an initial allocation policy unit. When the “details” button 1706 is pressed instead of the “apply” button 1705, the program 411 displays the portion of the reallocation log table 412 related to the information (the initial allocation policy) denoted in the check marked record.
The preceding has been an explanation of this embodiment.
According to this embodiment, even when the VVOL is used in a plurality of applications, an appropriate initial medium is determined for each application (combination of a virtual host WWN and a time period) just by defining the initial medium in VVOL units.
An embodiment of the present invention has been explained hereinabove, but the present invention is not limited to this embodiment, and it goes without saying that various changes can be made without departing from the gist thereof.
For example, in the embodiment, the initial medium is associated to both the virtual host WWN and the time period to which the write time belongs, but the initial medium may be associated with only one of these.
Further, the present invention is also applicable to a host for which host virtualization technology has not been applied, and in this case, the WWN becomes the host WWN. The identification information is not limited to the WWN, and other types of access-source identification information (for example, a port ID or IP address) may be used.
Further, for example, the initial allocation policy table 316 need not be used. In accordance with this, the storage control program 311 may determine a medium in accordance with a write each time there is a write, and may allocate a real area from the determined medium to the write-destination virtual area. Specifically, for example, the program 311 may determine the medium regarded as the initial medium based on another type of write-related information either besides or in addition to the write-source identification information and the write time, such as a write data attribute (for example, the size or format of the write data) or a write request attribute (an information element of the write request), may allocate a real area to this medium, and may write the write-targeted data to the allocated real area.
Further, in the above explanation, in a case where a real area is registered in a pool, the explanation was given on the premise that a RAID group is created, a logical volume is created from the RAID group, and the logical volume is registered in the pool. However, in a case where a different type of physical storage device is registered in the pool for providing the real area, another registration system may be used. For example, the following is conceivable.
Furthermore, as explained above, since this embodiment discloses the fact that the initial allocation tier of the real area is determined based on both the access-source identification information and the time period, the advantage of this embodiment is that a suitable initial allocation is realized for each application, or for each virtual host or for each host. However, the invention of the application may determine the initial allocation tier of the real area based on the access-source identification information without using the time period. This embodiment is significant in that a suitable initial allocation is realized for either each host or each virtual host, and is able to realize this initial allocation without acquiring the information inside either the virtual host or the host as shown in
Determining the initial allocation tier of the real area based on one time period is significant in that it makes it possible more detailed control than the access-source identification information in a case where one virtual host or host is used with a plurality of applications.
Furthermore, in the above explanation, a case in which a plurality of virtual hosts use a single virtual volume was explained, but the invention of this application is technology that is also applicable in a case where a plurality of applications or hosts other than a virtual host utilize a single virtual volume.
Number | Date | Country | Kind |
---|---|---|---|
2010-094059 | Apr 2010 | JP | national |