JP2008158733A - Cache control method in storage area network, switch device, and program - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide a cache control method, a switch device and a program of high locality of data, and light load of a cache operation without changing a host device and a storage device in a storage area network. <P>SOLUTION: A host device 2 transmits a data reading request including a destination address and a block address through a switch device 1 to a storage device XXX, and the storage device XXX transmits a data reading response including a block corresponding to the block address through the switch device 1 to the host device 2. Then, the switch device 1 transmits the transferred block from the storage device XXX to a storage device YYY to copy it. Then, when the host device 2 transmits the data reading request including the same block address as the block address to the switch device 1, the switch device 1 transmits the data reading request to the storage device YYY. <P>COPYRIGHT: (C)2008,JPO&INPIT

Description

  The present invention relates to a cache control method, a switch device, and a program in a storage area network (SAN).

  FIG. 1 is a system configuration diagram of a storage area network in the prior art.

  According to FIG. 1, a plurality of terminals 5 are connected to a plurality of host devices (servers) 2 via a LAN (Local Area Network) (IP (Internet Protocol) network). The LAN is composed of a plurality of LAN switch devices 4 and is usually connected by Ethernet (registered trademark). The LAN communicates using IP, and the LAN switch device 4 controls the route of the packet based on the IP header or the MAC header.

  The plurality of host devices 2 are connected to the plurality of storage devices 3 via the SAN. Thus, the storage device is connected to a dedicated network different from the LAN. The SAN is composed of a plurality of fiber channel switch devices 1 and is usually connected by optical fibers (or coaxial cables). For example, in the case of an optical fiber, a maximum transfer rate of 1.0625 Gbps (133 Mps in the case of a coaxial cable) is realized.

  As a prior art, there is a control device that manages a file stored in a storage device in association with an address in the storage device of the file (see, for example, Patent Document 1). When detecting a file read request from the terminal, the control device moves data in advance from the disk in the storage device to the cache memory in the storage device. This speeds up reading from the storage device.

  As another conventional technique, there is a technique in which the same data is stored in a plurality of storage devices, and a terminal can be connected to another storage device and read data in the same procedure (for example, see Patent Document 2). . This technique is realized by a control device provided in the storage device.

  Furthermore, a standard regarding a storage area network constituted by storage devices is also defined (for example, see Non-Patent Document 1).

JP 2005-148868 A JP 2006-309640 A Tetsuya Tsuji, “Where is the storage network going?”, @IT Information Management, [online], October 28, 2006, [Search on December 21, 2006], Internet <URL: http: // www .atmarkit.co.jp / im / cop / serial / storagenw / 06 / 01.html>

  However, according to the technique described in Patent Document 1, it is a cache control method in the same storage device and does not describe a technique for caching data among a plurality of storage devices. Further, according to the technique described in Patent Document 2, the control device in the storage device cooperates to back up a file to another storage device, and cooperation between different types of storage devices is not possible.

  Further, the cache control method in the prior art has a problem of disk capacity limitation because it is copied in file units. Another problem is that when a large-capacity file is cached, it affects the network in a burst manner.

  Therefore, the present invention provides a cache control method, a switch device, and a program for a storage area network that have a high data locality and a light cache operation load without changing the host device and the storage device. Objective.

According to the present invention, there is provided a cache control method in a switch device for connecting a host device and a plurality of storage devices via a storage area network,
The host device transmits a data read request including a destination address and a block address to the first storage device via the switch device, and the first storage device supports the block address to the host device via the switch device. A first step of transmitting a data read response including a block to be
A second step in which the switch device transfers the transferred block from the first storage device to the second storage device for copying;
A third step of transmitting a data read request to the second storage device when the host device transmits a data read request including the same block address as the block address to the switch device; Features.

According to another embodiment of the cache control method of the present invention,
A first routing table for the host device that stores the destination address of the storage device and the block address of the file stored in the storage device in association with each other;
A second routing table for the storage device that stores the address of the storage device and the block address of the file stored in the storage device in association with each other;
Each item of the first routing table is associated with each item of the second routing table,
Regarding the second step, it is also preferable that the address of the storage device related to the block in the second routing table is changed.

According to another embodiment of the cache control method of the present invention,
It is also preferable to transmit the routing table to another switch device and store the routing table received from the other switch device.

According to another embodiment of the cache control method of the present invention,
The switch device is provided in the Fiber Channel network,
The address of the storage device is specified by the Fiber Channel frame header,
The data read request is specified by a SCSI (Small Computer System Interface) command,
It is also preferable that the block address of the file is specified by the SCSI frame header.

According to the present invention, in a switch device that connects a host device and a plurality of storage devices via a storage area network,
A data read request received from the host device is transmitted to the first storage device using a routing table in which the address of the storage device and the block address of the file stored in the storage device are stored in association with each other. A routing means for transmitting a data read response including a block received from one storage device to the host device, and a copy for transferring the block transferred by the routing means to be copied from the first storage device to the second storage device Means,
Copy destination notifying means for notifying the routing means to change the address of the first storage device corresponding to the block address of the transferred block to the address of the second storage device.

According to another embodiment of the switch device of the present invention,
The routing table has a first routing table and a second routing table, and each item of the first routing table is associated with each item of the second routing table,
The first routing table associates a host device with a destination address of a storage device and a block address of a file stored in the storage device,
The second routing table associates the address of the storage device with the block address of the file stored in the storage device for the storage device,
It is also preferable that the address of the storage device related to the block in the second routing table is changed based on the information notified from the copy destination notification means.

  According to another embodiment of the switch device of the present invention, the routing means preferably transmits the routing table to the other switch device and accumulates the routing table received from the other switch device.

According to another embodiment of the switch device of the present invention,
Provided in the Fiber Channel network,
The address of the storage device is specified by the Fiber Channel frame header,
The data read request is specified by a SCSI command,
It is also preferable that the block address of the file is specified by the SCSI frame header.

According to the present invention, in a program for functioning a computer mounted on a switch device that connects a host device and a plurality of storage devices via a storage area network,
A data read request received from the host device is transmitted to the first storage device using a routing table in which the address of the storage device and the block address of the file stored in the storage device are stored in association with each other. A routing means for transmitting a data read response including a block received from one storage device to the host device, and a copy for transferring the block transferred by the routing means to be copied from the first storage device to the second storage device Means,
The computer is caused to function as a copy destination notifying unit that notifies the routing unit to change the address of the first storage device corresponding to the block address of the transferred block to the address of the second storage device.

  According to the cache control method, the switch device, and the program of the present invention, the data locality (cache performance) is high and the load of the cache operation is reduced without changing the host device and the storage device in the storage area network. be able to.

  Hereinafter, the best mode for carrying out the present invention will be described in detail with reference to the drawings.

  According to the present invention, one data is stored in a SAN by a plurality of storage devices. This has not only a role of data backup but also a role of realizing high-speed data reading in response to a data reading request from the host device. For example, the host device can cache data read in the past or data that is highly likely to be read in a storage device that is physically or logically close to the host device.

  FIG. 2 is a sequence diagram of SAN in the present invention.

  First, it is assumed that the storage device XXX stores a predetermined file. This file is divided into four blocks, and block addresses 1 to 4 are given. Further, it is assumed that the position of the storage device YYY is physically or logically closer than the position of the storage device XXX when viewed from the host device 2. Therefore, the host device 2 can execute data reading from the storage device YYY at a higher speed than reading data from the storage device XXX.

(S201) The host device 2 transmits a data read request to the switch device 1. This data read request is a request to read the block at block addresses 2 and 3 for the file stored in the storage device XXX. This data read request uses the destination address as the address of the storage device XXX.

(S202) The switch device 1 transfers the data read request received from the host device 2 to the storage device XXX based on the destination address (address of the storage device XXX).

(S203) In response to the data read request, the storage device XXX reads the previously stored blocks 2 and 3, and returns a data read response to which these blocks have been added to the switch device 1.

(S204) The switch device 1 transmits the data read response received from the storage device XXX to the host device 2.
(S205) At this time (or thereafter), the switch device 1 transmits a data write request to the storage device YYY closer to the storage device XXX when viewed from the host device 2. The blocks 2 and 3 received from the storage device XXX are added to the data write request.
(S206) The storage device YYY receives the data write request from the switch device 1, and stores the added block in the storage device YYY. Then, the storage device YYY returns a data write response to the switch device 1.

(S207) Thereafter, the host device 2 transmits a data read request to the switch device 1. This data read request is for requesting reading of the blocks at block addresses 2 and 3 as before. This data read request uses the destination address as the address of the storage device XXX.
(S208) At this time, the switching device 1 transfers the data read request received from the host device 2 to the storage device YYY.

(S209) In response to the data read request, the storage device YYY reads the data of the blocks 1 and 2 stored in advance, and returns a data read response to which these blocks are added to the switch device 1.
(S210) The switch device 1 transmits the data read response received from the storage device YYY to the host device 2. As a result, the blocks at the block addresses 2 and 3 of the file are read from the storage device YYY. This can be realized at higher speed than reading from the storage device XXX. Further, the host device and the storage device are existing and need not be changed at all.

  FIG. 3 is a first functional configuration diagram of the switch device according to the present invention.

  According to FIG. 3, the switch device 1 includes a routing unit 11, a copy unit 12, a copy destination notification unit 13, and a port 14. Functional units other than the port 14 can also be realized by executing a program that causes a computer mounted on the switch device 1 to function.

  The routing unit 11 has routing tables 1 and 2 and performs frame path control. The frame received from the first port 14 is transmitted to the second port 14 based on the routing tables 1 and 2.

  The routing table 1 stores the destination address of the storage device and the block address of the file stored in the storage device in association with each other in the host device. The routing table 1 includes a host address, a destination address, a LUN (Logical Unit Number), and a block address. The host address is different for each host device. The destination address is the address of the storage device in which the block is stored. The host device first knows the address of the storage device. LUN represents a disk number in the storage device. The block address represents an address of a block unit stored in the disk of the LUN in the storage device. Thereby, data can be read or written in units of blocks.

  The routing table 2 stores a storage device address and a block address of a file stored in the storage device in association with each other. The routing table 2 includes a storage device address, a storage device LUN, and a block address. The storage device address is an address that can be controlled by the switch device. The storage device LUN represents a disk number in the storage device. The block address represents the address of a block stored in the disk of the LUN in the storage device.

  The routing tables 1 and 2 shown in FIG. 3 are associated one-to-one with each item (block unit). Thereby, it is possible to grasp to which storage device the data read request should be transmitted in block units.

  The switch device 1 can also transmit the routing tables 1 and 2 to other switch devices, and can accumulate the routing tables 1 and 2 received from other switch devices. As described above, by exchanging the routing table between the neighboring switch devices 1, the same block stored in the storage device physically or logically located nearby can be easily found.

  The copy unit 12 copies the data stored in the first storage device to the second storage device. Such a copy has not only a role of data backup but also a role of realizing high-speed data reading in response to a data reading request from the host device. It is stored in block units in a storage device that is physically or logically close to the host device that has transmitted the data read request.

  The copy destination notifying unit 12 notifies the routing unit 11 of the copy operation by the copy unit 12. The routing unit 11 is notified to change the address of the first storage device corresponding to the block address of the copied block to the address of the second storage device.

  Next, it demonstrates along the sequence of FIG.

  The host device 2 is connected to the storage device XXX and the storage device YYY via the switch device 1. First, it is assumed that the storage device XXX stores a predetermined file. This file is divided into four blocks, and block addresses 1 to 4 are given. Further, it is assumed that the position of the storage device YYY is physically or logically closer than the position of the storage device XXX when viewed from the host device 2. Therefore, the host device 2 can execute data reading from the storage device YYY at a higher speed than reading data from the storage device XXX.

(S301) The switch device 1 receives a data read request from the host device 2. In the data read request, the destination address is the storage device XXX, and the block addresses are 2 and 3. The routing table 1 in the routing unit 11 of the switch device 1 is initially routed to the storage device XXX for all blocks (block addresses 1 to 4). Therefore, the routing unit 11 transmits this data read request to the storage device XXX.

  On the other hand, the switch device 1 receives the data read response including the blocks 2 and 3 from the storage device XXX. This data read response is transferred to the host device 2 by the routing unit 11.

(S302) The routing unit 11 notifies the copy unit 12 that the blocks “2 and 3” stored in the storage device XXX have been read by the host device 2. As a result, the copy unit 12 transfers the blocks at the block addresses 2 and 3 stored in the storage device XXX to be copied to the storage device YYY. With this operation, the storage device YYY stores the blocks 2 and 3 that have been previously read by the host device 2.

(S303) The copy destination notification unit 13 notifies the routing unit 11 of the copy operation by the copy unit 12. According to FIG. 3, the copy destination notifying unit 12 notifies the routing unit 11 of information indicating that the blocks having the block addresses 2 and 3 have been copied from the storage device XXX to the storage device YYY.

  As a result, the routing unit 11 rewrites the routing table 2. According to the routing table 2 after copying in FIG. 3, it can be understood that the block addresses 2 and 3 of the destination address XXX in the routing table 1 are stored as the block addresses 1 and 2 in the storage device YYY.

(S304) It is assumed that the switch device 1 receives a data read request for the blocks 2 and 3 from the host device 2 again. Similarly, in the data read request, the destination address is the storage device XXX, and the read block addresses are 2 and 3. At this time, the routing table 2 in the routing unit 11 is route-controlled for the block addresses 2 and 3 to the storage device YYY. Therefore, the routing unit 11 converts the data read request into block addresses 1 and 2 and transmits the block addresses to the storage device YYY.

  In response to this, the storage device YYY returns a data read response including the blocks 1 and 2 to the switch device 1. The data read response is transferred to the host device 2 by the routing unit 11.

  FIG. 4 is a second functional configuration diagram of the switch device according to the present invention.

  The functional configuration of FIG. 4 is different from that of FIG. 3 in that the copy unit 12 is provided between the routing unit 11 and the port 14. According to the configuration in FIG. 4, the routing unit 11 does not need to notify the copy unit 12 of the read block address. The copy unit 12 can monitor the copy operation of the routing unit 11.

(S401) The operation is the same as S301 in FIG. However, the routing unit 11 is different in that it communicates with the storage device XXX connected to the port X via the copy unit 12.

(S402) The copy unit 12 knows that the blocks “2 and 3” stored in the storage device XXX have been read by the host device 2. As a result, the copy unit 12 transfers the blocks at the block addresses 2 and 3 stored in the storage device XXX to be copied to the storage device YYY. With this operation, the storage device YYY stores the blocks 2 and 3 that have been previously read by the host device 2.

(S403) The operation is exactly the same as S303 in FIG.
(S404) The operation is the same as S304 in FIG.

  FIG. 5 is an explanatory diagram showing the transition of the routing table in the present invention.

  The routing tables shown in FIGS. 3 and 4 correspond one to one for each block address. On the other hand, in the routing table of FIG. 5, the storage device is specified by the block range.

(S501) The blocks having the block addresses 1 to 30000 are stored from the block address 1 in the storage device XXX.

(S502) It is assumed that the host device reads a block having block addresses 10001 to 15000 from the storage device XXX. At this time, the routing table 1 is divided into three block addresses 1 to 10,000, 10001 to 15000, and 15001 to 30000. According to the routing table 2, the block addresses 10001 to 15000 are stored from the block address 1 in the storage device YYY.

(S503) Next, it is assumed that the host device reads a block having block addresses 20001 to 30000 from the storage device XXX. At this time, the routing table 1 is divided into four block addresses 1 to 10,000, 10001 to 15000, 15001 to 20000, and 20001 to 30000. According to the routing table 2, the block addresses 20001 to 30000 are stored in the storage device YYY from the block address 5000.

(S504) Next, it is assumed that the host device reads a block having block addresses 15001 to 20000 from the storage device XXX. At this time, according to the routing table 2, the block addresses 15001 to 20000 are stored in the storage device YYY from the block address 15001.

  FIG. 6 is a frame configuration diagram of the SAN protocol.

  The SAN protocol is composed of Fiber Channel and SCSI. The switch device controls the frame path based on the Fiber Channel protocol header. Further, reading and writing of data stored in the storage device is controlled by a SCSI command.

  According to FIG. 6, the fiber channel frame includes a fiber channel header and a SCSI command part. The fiber channel header includes a destination address and a source address. In the case of a request command, the address of the storage device is included in the destination address, and the address of the host device is included in the source address.

  The SCSI command part includes a LUN, an operation code, and a block address. LUN is the disk number of the storage device. The operation code includes, for example, a data read request, a data read response, a data write request, and a data write response. The block address represents the address of the block on the disk of the storage device.

  As described above in detail, according to the cache control method, the switch device, and the program according to the present invention, the storage area network is not subjected to load or change on the host device and the storage device, and the local data is not changed. (Cache performance) can be improved.

  In particular, according to the present invention, the switch device can detect a SCSI command, and controls the path in units of blocks according to the contents thereof. In addition, the cache operation is automatically performed by the data read operation and does not require a separate cache operation request.

  According to the various embodiments of the present invention described above, those skilled in the art can easily make various changes, modifications and omissions within the scope of the technical idea and the viewpoint of the present invention. The above description is merely an example, and is not intended to be restrictive. The invention is limited only as defined in the following claims and the equivalents thereto.

It is a system configuration | structure figure of the storage area network in a prior art. It is a sequence diagram of SAN in the present invention. It is a 1st functional block diagram of the switch apparatus in this invention. It is a 2nd function block diagram of the switch apparatus in this invention. It is explanatory drawing showing transition of the routing table in this invention. FIG. 3 is a frame configuration diagram of a SAN protocol.

Explanation of symbols

DESCRIPTION OF SYMBOLS 1 Switch apparatus, Fiber Channel switch apparatus 11 Routing part 12 Copy part 13 Copy destination notification part 14 Port part 2 Host apparatus 3 Storage device 4 LAN switch apparatus 5 Terminal

Claims (9)

A cache control method in a switch device that connects a host device and a plurality of storage devices via a storage area network,
The host device transmits a data read request including a destination address and a block address to the first storage device via the switch device, and the first storage device transmits to the host device via the switch device. A first step of transmitting a data read response including a block corresponding to the block address;
A second step in which the switch device transfers the transferred block from a first storage device to a second storage device for copying;
When the host device transmits a data read request including the same block address as the block address to the switch device, the switch device transmits the data read request to a second storage device. And a cache control method. A first routing table for the host device that stores a destination address of the storage device and a block address of the file stored in the storage device in association with each other;
A second routing table for the storage device that stores the address of the storage device and the block address of the file stored in the storage device in association with each other;
Each item of the first routing table is associated with each item of the second routing table,
The cache control method according to claim 1, wherein the address of the storage device related to the block in the second routing table is changed in the second step.   3. The cache control method according to claim 2, wherein the routing table is transmitted to another switch device and the routing table received from the other switch device is accumulated. The switch device is provided in a fiber channel network,
The address of the storage device is specified by a Fiber Channel frame header,
The data read request is specified by a SCSI (Small Computer System Interface) command,
The cache control method according to any one of claims 1 to 3, wherein the block address of the file is specified by a SCSI frame header. In a switch device that connects a host device and a plurality of storage devices via a storage area network,
A data read request received from the host device is transmitted to the first storage device using a routing table in which the address of the storage device and the block address of the file stored in the storage device are stored in association with each other. A routing unit that transmits a data read response including a block received from the first storage device to the host device; and the block transferred by the routing unit is copied from the first storage device to the second storage device. Copy means to transfer as much as possible,
Copy destination notifying means for notifying the routing means to change the address of the first storage device corresponding to the block address of the transferred block to the address of the second storage device. Switch device. The routing table has a first routing table and a second routing table, and each item of the first routing table is associated with each item of the second routing table,
The first routing table associates the destination address of the storage device and the block address of the file stored in the storage device with respect to the host device,
The second routing table associates the address of the storage device with the block address of the file stored in the storage device for the storage device,
6. The switch device according to claim 5, wherein an address of the storage device related to the block in the second routing table is changed based on information notified from the copy destination notifying unit.   The cache control method according to claim 6, wherein the routing unit transmits the routing table to another switch device and accumulates the routing table received from the other switch device. Provided in the Fiber Channel network,
The address of the storage device is specified by a Fiber Channel frame header,
The data read request is specified by a SCSI command,
The switch device according to claim 5, wherein the block address of the file is specified by a SCSI frame header. In a program for functioning a computer mounted on a switch device that connects a host device and a plurality of storage devices via a storage area network,
A data read request received from the host device is transmitted to the first storage device using a routing table in which the address of the storage device and the block address of the file stored in the storage device are stored in association with each other. A routing unit that transmits a data read response including a block received from the first storage device to the host device; and the block transferred by the routing unit is copied from the first storage device to the second storage device. Copy means to transfer as much as possible,
The computer is caused to function as a copy destination notifying unit that notifies the routing unit to change the address of the first storage device corresponding to the block address of the transferred block to the address of the second storage device. A program for a switch device.

Images