KR100541530B1 - Method of processing snapshot introducing snapshot status table based on memory for shared storage supporting large capacity - Google Patents

Abstract

The present invention relates to a snapshot technique for a mass storage volume based on a mapping in an environment in which mass storage devices are shared by multiple hosts, and the snapshot state table (SST) of the present invention is stored on a memory rather than a disk. It consists of entries consisting of Data-block Allocation-time Bits (DABs) and COW of Data-block Bits (CDBs), which indicate whether to perform copy-on-write (COW), which indicates when to allocate a data block. do.

Therefore, in the present invention, additional disk access is not required for determining whether to perform COW when performing a write operation and a snapshot deletion on the volume where the snapshot exists. It is possible to improve the performance of the snapshot technique by eliminating the access to the mapping block by completely removing the access to the mapping block by using the SST in memory to determine whether the COW has been performed using the mapping entry of the original mapping block stored in the disk. have.

Snapshot, mapping table, snapshot state table, copy-on-write

Description

Method of processing snapshot introducing snapshot status table based on memory for shared storage supporting large capacity}

1 illustrates a copy-on-write (COW) process of a general snapshot technique.

2 is a diagram illustrating a configuration environment of a multi-host system sharing storage to which the present invention is applied.

Figure 3 illustrates the structure of a snapshot status table and entries in accordance with the present invention.

4A and 4B are flowcharts illustrating an operation of deleting a snapshot according to an embodiment of the present invention.

5A and 5B are operational flowcharts of a write operation according to an embodiment of the present invention.

<Description of the symbols for the main parts of the drawings>

200: storage area network environment 202: network storage device

201: switch 203: host

204: LAN 110: original file system

120: Snapshot area 301: Snapshot state table

302: Snapshot State Table Entry 303: Logical Address

304: Data Block Allocation Bits (DAB) 305: COW Data Block Bits (CDB)

The present invention relates to a snapshot technique for a mass storage device in a shared storage environment, and more particularly, to perform a snapshot of a mass storage device in a shared storage cluster environment based on a storage area network (SAN). It is about a method.

In enterprise-scale systems, the amount of data shared and serviced to users is growing exponentially with the establishment of data warehouses, enterprise resource planning (ERP), and customer relationship management (CRM) systems. In other words, due to the explosive growth of the Internet and the necessity of maintaining information online, the need for collecting and tracking decision support information, PC server movement toward server consolidation and business-oriented applications, and increased application complexity The demand for storage continues to rise. It is the Storage Area Network (SAN) that emerged as a solution to meet high-speed, high-capacity data processing needs. SANs are becoming mainstream in distributed networking, and they are already the basic conditions for configuring storage.

On the other hand, enterprise systems that need to support 24 x 7 x 365 environments require data availability and reliability in addition to high-speed, high-volume data processing. Among the many ways to ensure reliability and availability, the importance of online backup is emerging as a way to meet the requirements of these systems. However, in order to perform a backup of a large amount of data, the backup execution time increases exponentially. Therefore, it is essential to provide an online snapshot based on a mapping table that can perform backup without interruption of service or restart of the system.

However, there are some problems when applying the mapping table-based snapshot technique to a large capacity shared storage system based on SAN.

First, copy-on-write (COW), which handles the initial changes to the data blocks after the snapshot is taken, and changes to the data blocks that occur after the COW are additional disks to the mapping block to determine whether to perform the COW. Input / output (I / O) is required, resulting in poor performance of write operations.

Second, even when the snapshot deletion is performed, additional disk input / output (I / O) is required for the mapping block in the process of determining whether the data block is COW performed when deallocating the data block. There is a problem in that performance degradation occurs, and the time for performing the snapshot deletion is long.

1 is a schematic diagram illustrating a process of performing Copy-on-Write (COW) in a general disk-based snapshot technique.

Referring to FIG. 1, reference numeral 110 denotes an original file system having the most recent data, and 120 denotes a snapshot for storing data necessary to maintain a snapshot state for changed data when a snapshot is created. Area. In the file system 110, a file A 111 and a file B 112 are stored. The file A 111 is composed of A1 and A2 blocks, and the file B 112 is composed of B1, B2 and B3 blocks. have.

In this structure, when a read request for the first block A1 of file A 111 is received before a snapshot is generated from an application program (151), file A (stored in file system 110) is stored. Read the block A1 of 111) to perform the service.

However, if a write to the first block B1 of the file B 112 is requested 154 after the snapshot is created, a process 140 for checking whether it is the first update is necessary. In other words, it is necessary to check whether the COW has been performed. If the write operation for the first block B1 of the file B 112 is the first update, a copy on write COW should be performed.

The copy on write process copies the first block B1 of the file B 112 in the file system 110 to the blocks B1 ': 122 of the newly allocated snapshot area 120. The mapping information for the snapshot data block is recorded in the changed block map 121. The actual update contents are recorded in the first block B1 of the file B 112 in the original file system 110. If it is not the first update, the changed data is recorded in the first block B1 of the file B 112 of the original file system 110.

Meanwhile, even when performing a read operation on the snapshot, a process 130 for checking whether the block has been changed by the copy on write COW is required. If the block is not changed, the data of the original file system 110 is immediately read. However, if a change occurs, the changed block map 121 of the snapshot area 120 is inspected to find the physical location 122 of the changed block and read the data. That is, if it is a read 153 of the first block A1 of file A, it is an unchanged block, so that block A1 of the original file system 110 is read, and a read of the first block B1 of file B is performed. In step 152, the block B1 'of the snapshot area is read because it is a changed block.

As described above, additional input / output (I / O) to the disk in which the mapping block is stored to access the mapping block to check whether the COW is performed when the snapshot is read or the snapshot is deleted is conventionally performed. There is a problem in that performance of a write operation that is requested and executed concurrently occurs, and a snapshot deletion execution time becomes long.

Accordingly, an object of the present invention is to provide a method for performing a performance-enhanced snapshot for a large-capacity logical volume on a storage area network (SAN) that provides large shared storage. have. In other words, the present invention maintains a table indicating a snapshot status in memory to read a mapping block of a disk that is required for a write operation when a snapshot exists and the determination of a copy on write when the snapshot is deleted. To improve performance and save snapshot state table entries to disk with initial data-block allocation location (DAB) and COW of snapshot bit (CDB) values that indicate this after COW It does not require a process to improve the performance of write operations.

In order to achieve the above object, the snapshot deletion method of the present invention provides a method of deleting a snapshot in a storage area network (SAN) -based system in which a plurality of hosts share a large amount of network storage. A first step of changing an operation mode of all nodes in which a server exists to a snapshot deletion (SNAPSHOT_DESTROY) mode; Obtaining a lock on the mapping block; A third step of processing whether to perform copy-on-write (COW) of the data block indicated by the mapping entry by using the first allocation bit (DAB) and the COW data block bit (CDB) of the snapshot state table entry; A fourth step of initializing a first allocation bit (DAB) or a COW data block bit (CDB) when the copy-on-write (COW) is performed and unlocking the mapping block; And a fifth step of deleting a snapshot when initialization of all data blocks is completed.

In addition, in order to achieve the above object, a snapshot performing method of the present invention provides a method for performing a write operation in a storage area network (SAN) -based system in which a plurality of hosts share a large amount of network storage. Determining whether a shot exists or not and performing mapping; A second step of reading the mapping block from a disk to obtain a value of a desired mapping entry; Checking a data block allocation bit (DAB) value of a snapshot state table entry to determine whether a data block is first allocated and used after a snapshot is created; A fourth step of allocating a new block, writing the data contents to a copy disk, and changing the datablock allocation bit (DAB) value to 1 when the datablock allocation bit (DAB) value is 0 and the mapping entry value is an initial value. ; A fifth step of determining whether a copy on write has been performed through a COW data block bit (CDB) value when the data block has been allocated before the snapshot; And a sixth step of performing a copy on write (COW) and changing the COW data block bit (CDB) value to 1 when the COW data block bit (CDB) value is 0 and COW is not performed. It is done.

Hereinafter, exemplary embodiments of the present invention will be described with reference to the accompanying drawings.

In an embodiment of the present invention, a snapshot technique for mass shared storage is applied to a Logical Volume Manager (LVM). In addition, the algorithm of the embodiment according to the present invention will be described by dividing into a snapshot delete and a write operation.

FIG. 2 is a system configuration diagram illustrating a SAN-based physical environment providing mass shared storage between multiple hosts according to the present invention, wherein a logical volume manager (LVM) stores storage in a multi-host environment through a storage area network (SAN). A shared storage cluster system is provided that provides sharing.

The SAN environment 200 on which the Logical Volume Manager (LVM) is based includes a Network Storage Pool (202) for storing and storing data, and the Network Storage Pool (202), as shown in FIG. The multiple hosts (Host1-Host4: 203), and the network storage pools (202) and hosts (203) via Fiber Channel (FC) to share them. It is composed of switches (201) which is a storage network network. Here, the network storage pool 202 is a storage device consisting of a backup device, a disk array, a disk, a RAID, a JBOD, and the like.

In this configuration, each host 203 uses a host bus adapter (HBA) for connection with the switch 201, and the connection between the host 203 and the switch 201 uses a Fiber Channel (FC). Is configured using. Each host (Host1-Host4) 203 is connected to a LAN 204 to process control and management information. When this physical environment is ready, you are ready to use Logical Volume Manager (LVM).

3 shows an entry structure of a snapshot state table for implementing a snapshot proposed in the present invention.

Referring to Fig. 3, the snapshot status table 301 of the present invention is implemented in main memory, and each entry 302 of the snapshot status table is composed of bits representing two kinds of snapshot status information. At this time, each entry 302 corresponds to a logical address (0 to N: 303) of the mapping block entry. That is, one snapshot state table entry 302 is assigned to one mapping block entry. Each entry 302 is made up of data block allocation bits (DAB) 304 and COW data block bits (CDB) 305.

The datablock allocation bit (DAB) 304 allocates and uses the first bit of every snapshot state table entry 302. The DAB 304 is a bit used to distinguish the data block that is first allocated and used after the snapshot is taken. Change the DAB value to '1' when the data block is allocated and used for the first time after taking the snapshot. The COW data block bit (CDB) 305 is a bit indicating the status of the snapshot. If '1', it means that the COW was performed after the snapshot was created. Indicates that the COW has not been performed. The CDB 305 is allocated one bit per snapshot, and as many bits as the maximum number of snapshots are allocated. The data block allocation bit (DAB) 304 and the COW data block bit (CDB) 305 are initialized to '0' during the creation and initialization of the snapshot state table 301 in memory when the snapshot is first created.

Typically, disk-based snapshot techniques use both the original mapping snapshot entry and the snapshot mapping entry to determine if subsequent data change operations must perform copy on write (COW) for the first block used after the snapshot. Read and compare the values of the physical addresses. In other words, two disk I / Os must be performed. In addition, in the process of returning the allocation of the copy on write (COW) execution block performed at the time of deleting the snapshot, the return operation of the data block should not be performed for the block first allocated after the creation of the snapshot. To make this determination, the existing snapshot technique requires two disk input / output (I / O) of the original mapping entry and the snapshot mapping entry. In addition, a method of maintaining and processing COW data block bits (CDB: 305) in the original mapping table of the disk requires disk input / output (I / O) as access to the original mapping block.

However, in the snapshot scheme according to the present invention, the data block allocation bit (DAB) 304 and the COW data block bit (CDB) 305 of the snapshot state table entry 302 stored in the memory are used to map the disk to the mapping block. It is possible to determine whether to perform a copy on light (COW) without performing I / O for each other. To this end, the DAB 304 value of the snapshot state table entry is changed to '1' for the first data block allocated / used after the snapshot is created.

Accordingly, in the present invention, if the DAB 304 of the corresponding entry is '1' in the process of determining whether to perform a copy on write (COW), the contents of the data block may be changed and the change may be reflected in the disk as it is. If the DAB 304 of the entry is '0' as an already allocated block, it is determined through the COW data block bit (CDB) 305 value.

When a snapshot exists, each data block can be classified into three cases according to its state.

The first is the data block that is allocated and used for the first time after the snapshot is created. The second is a block of data that is used before the snapshot and has not yet changed since the snapshot. That is, the data block is not performed COW. The third is the data block where the change occurred after the snapshot, that is, the COW was performed. In general, the state of a data block that is changed after a snapshot is created is one of the above three, and in the second case, a COW should be performed when a change occurs for the data block.

As in the first case, there may be data blocks first allocated after the snapshot is created. When a change operation on such a data block occurs, the first change allocates a new data block and writes to the disk. However, COW is not performed because the block was not used before the snapshot. That is, since the COW has not been performed, the value of the COW data block bit (CDB: 305) is kept at the initial value '0'. However, even if it is the first allocated data block, COW is performed because the data block is in use and the CDB value is '0' from the second change. That is, the snapshot technique according to the present invention does not perform additional disk access to read mapping entries recorded on the disk in the process of determining whether to perform COW. Therefore, the first used block after the snapshot is created and allocated before the snapshot is created. It is impossible to distinguish blocks that are not COW performed. Therefore, in the present invention, an additional 1 bit (this is "DAB") other than the CDB 305 is displayed to indicate a data block allocated and used for the first time after the snapshot is created, and after the snapshot is generated by the DAB 304. Distinguish between blocks that are first used and blocks that were allocated before the snapshot was taken and no COW was performed. In the third case, the value of the CDB 305 has already been changed to '1' by the COW, and only a write operation on the original data block needs to be performed even if a change operation occurs.

1. Delete Snapshot

One of the factors causing performance degradation in the conventional snapshot technique is the performance degradation due to the overhead of accessing the snapshot mapping block to determine whether to perform COW when deallocating data blocks during the snapshot delete operation. . That is, in the conventional snapshot technique, both the original mapping entry and the snapshot mapping entry are read to compare whether the COW has been performed, and then the addresses of the physical blocks pointed to by the two entries are compared. If you have more than one snapshot, you need to compare mapping entries from other snapshots as well. That is, at least two additional disk I / O operations must be performed.

As described above, in order to solve the problem of a snapshot delete operation in which performance is degraded due to the access to the mapping block, the present invention allocates a COW data block bit (CDB) 305 and a data block to each entry 302 of the snapshot state table. This is solved by introducing a snapshot state table 301 having a bit (DAB) 304 value into the main memory.

As described above, in the snapshot technique according to the present invention, it is possible to determine whether to perform COW using the entry 302 of the snapshot state table without using a method of determining using the value of the mapping block of the disk.

First, the value of the data block allocation bit (DAB: 304) of the snapshot state table entry is checked, and if the data block allocation bit (DAB: 304) is '1', it is checked whether the snapshot to be deleted is the first snapshot. . If the DAB 304 is '1' and is the first snapshot, the corresponding data block is the first allocated block after the snapshot is created and thus is not deallocated. Otherwise, the COW data block bit (CDB: 305) value is checked. If the CDB 305 value of the snapshot position to be deleted is '1', it is a case where a copy on write (COW) is performed. If the COW data block bit (CDB: 305) is '0', execution continues for the next data block. If the CDB 305 is '1', the operation checks whether the next snapshot exists and if the snapshot exists. The COW data block bit (CDB: 305) value of the snapshot state table entry 302 is checked. If the next snapshot does not exist or if the snapshot exists and COW is performed, the current data block must be deallocated. In this case, when the next snapshot exists, it is determined whether to perform the COW through the CDB 305 value to determine whether to release the allocation to the current snapshot.

Figure 4 is a flow chart illustrating the flow of snapshot deletion according to the present invention, look at the flow of snapshot deletion of the present invention as follows.

When a snapshot deletion is requested, the volume operation mode of the mapping server is changed to a "snapshot deletion (SNAPSHOT_DESTROY)" mode (401). The reason for changing to the snapshot deletion mode is to prevent the COW for the snapshot on the data block occurring before the snapshot deletion is completed. A location of the disk block in which the mapping block is stored is obtained (402), and an exclusive mode lock on the mapping block is obtained (403).

The mapping block is read from disk to obtain a mapping entry (404). The data block allocation bit (DAB: 304) and the COW data block bit (CDB: 305) values of the snapshot state table entry should be checked to determine whether the data block is deallocated. A check is performed on entries 302 of all snapshot status tables in order to check whether COW is performed.

It is determined whether a copy on write (COW) has been performed through the values of the DAB 304 and the CDB 305 of the snapshot state table entry (405). If the block has not been COWed, the next snapshot state table entry is checked (409).

If the COW has been performed, the operation of determining whether to deallocate the data block is performed (406). That is, there are two cases in which the allocation of data blocks on which COW is performed is deallocated. First, when the next snapshot does not exist (next snapshot = null), and second, when the next snapshot exists, COW occurs in the same data block after creating the next snapshot. In both cases, the allocated data block is released (407), and the CDB 305, which is the status bit value for the current snapshot of the snapshot status table entry, is initialized to '0' (408). Check 409 if all snapshot status table entries 302 have completed. If not, then the operations from step 405 to step 408 are performed for the next snapshot state table entry.

Then, the lock on the mapping block is released (410). After unlocking, it is checked whether execution is completed for all mapping blocks of the volume (411). If there is a mapping block to be performed, the next mapping block is obtained in step 402, and the process up to step 410 is performed.

If all mapping blocks have been performed, the operation of deleting the snapshot from the host 203 is performed (412). When the snapshot is deleted, the actual snapshot deletion is completed, and the configuration information of the original volume changed by the snapshot deletion is reflected (413). Finally, when the operation mode of the mapping server is converted to normal mode (414), the snapshot deletion process is completed.

2. Data write operation

When a snapshot exists, a write operation that reflects a change in a data block requires a process of determining whether to perform COW as in the case of deleting the snapshot. In order to make this determination, the existing snapshot method based on the mapping table performs the decision by accessing the mapping block, and the access to the mapping block degrades the performance of the write operation.

The contents of data written before the snapshot was taken should be retained even if changes occur. This operation to maintain the contents of the data block allocated before the snapshot is called copy-on-write (COW). This COW process must be performed in the same way in the snapshot technique based on the mapping table. The problem is the handling of data block changes that occur after COW is performed. The existing snapshot technique reads both the original mapping entry and the snapshot mapping entry to determine whether the data block intended to reflect the change has been performed, and compares the address of the physical block that the two entries map to to determine whether COW is performed. Judged.

In the present invention, the write operation is solved by using the DAB 304 and the CDB 305 of the snapshot state table in the same manner as in deleting the snapshot. Initially, the DAB 304 value is checked to see if a COW has occurred for the first allocated data block after snapshot creation. If the DAB 304 value is '0', the CDB 305 value is checked. The CDB 305 is changed to '1' when the contents change of the data block first occurs after the snapshot is generated for the data block allocated and used before the snapshot. That is, the value of the CDB 305 corresponding to the snapshot of the snapshot state table entry 302 where COW is performed is changed to '1'. When a change to the same data block occurs after the COW is performed, it is determined whether the COW is performed through the CDB 305 value of the snapshot state table entry. That is, the snapshot technique according to the present invention does not require reading the snapshot mapping entry and comparing the original mapping entry with the original mapping entry, and does not need to read the value of the original volume mapping entry. That is, determination can be made using the snapshot state table entry 302 in memory without disk access to the mapping block.

5 is a flowchart of a write operation for processing a content change request of a disk block of a user when a snapshot exists in the system according to the present invention.

First, a physical address of a mapping block in which mapping information about a data block is stored is obtained (501). An exclusive mode lock is acquired on the mapping block (502), and the mapping block is read from the disk to obtain a mapping entry corresponding to the logical address (503). The operation mode of the volume is checked (504). That is, it is checked whether the current I / O occurred during snapshot creation (SNAPSHOT_CREATE) or deletion (SNAPSHOT_DESTROY). If the operation mode is NORMAL, it is checked whether a snapshot exists (505). If the snapshot does not exist, the data block is written to disk as in a normal write operation (516), and the mapping block is unlocked and terminated (517).

If a snapshot exists in step 505, it is checked whether data blocks are allocated (507). If the data block is not allocated, the data block is allocated (520), and the physical block address of the mapping entry is changed to the newly allocated data block (521). When the value of the mapping entry is changed, the DAB 304 value of the snapshot state table entry is changed to "1" (522). The mapping block is then written to the disk (523). The changed contents are recorded on the disk of the data block (516), the lock on the mapping block is released (517), and the process ends.

If the data block has been allocated in step 507, it is checked if the allocation has been made after snapshot creation (508). COW is not performed on the data blocks allocated after the snapshot is created. If allocated prior to snapshot creation, it is checked if the COW has been performed using the CDB 305 value of the snapshot status table entry (509). If the COW has already been performed as a result of the check, the data block is written to the disk (516), the lock on the mapping block is released, and the execution ends (517).

If the COW has not yet been performed in step 509, the COW should be performed. To this end, the snapshot mapping block is read into the buffer to obtain a snapshot mapping entry (510), a new physical data block is allocated to perform a copy on write (511), and the contents of the data block are newly allocated. The data block is copied and the copied data block is written to the disk (512).

Subsequently, the physical address indicated by the snapshot mapping entry is changed to the address of the newly allocated data block (513), the CDB 305 value of the snapshot state table entry indicating that the COW has been performed is changed to "1" (514). The snapshot mapping block is written to the disk (515). When the COW process from step 510 to step 515 is performed, the contents of the data block are written to the disk (516), the lock on the mapping block is terminated, and the execution is terminated (517).

If the operation mode is not the normal mode and the snapshot creation (SNAPSHOT_CREATE) mode in step 504, it is checked whether copying to the mapping data block including the mapping entry has been performed (506). When the copy is completed, the CDB 305 value of the snapshot status table is checked to determine whether the COW has been performed (509). If the CDB value is '1', the data block is written to the disk when COW has already been performed (516), the lock on the mapping block is released (517), and the execution ends. If the CDB value of the snapshot status table is '0', COW operation should be performed. Accordingly, the COW operation from step 510 to step 515 is performed, the data block is written to the disk (516), the mapping block is unlocked (517), and the execution ends. If the copy has not been completed in step 506, COW should be performed. The operation from step 510 to step 517 is performed and ends.

The snapshot performing method of the mass shared storage device using the memory-based snapshot status table of the present invention as described above may be stored in a computer-readable recording medium. Such a recording medium includes all kinds of recording media in which programs and data are stored so that they can be read by a computer system. Examples of the recording medium include read only memory, random access memory, and compact disk. -Rom, DVD (Digital Video Disk) -Rom, magnetic tape, floppy disk, optical data storage device. In addition, these recording media can be distributed over network coupled computer systems so that the computer readable code is stored and executed in a distributed fashion.

As described above, the large-capacity snapshot performing method according to the present invention improves the performance of write operations in a large shared storage environment, and is home trading, which is a real-time stock information and trading service through e-commerce or the web, which requires service and business continuity. To ensure high data availability and reliability in enterprise systems such as In addition, the present invention has the advantage that can support the online backup without the performance degradation of the write operation.                     

What has been described above is only one embodiment for explaining an efficient snapshot performing method for a mass shared storage device according to the present invention, and the present invention is not limited to the above-described embodiment, and is claimed in the following claims. Without departing from the gist of the present invention, any person having ordinary knowledge in the field of the present invention will have the technical idea of the present invention to the extent that various modifications can be made.

Claims (10)

In the method of performing a snapshot in a storage area network (SAN) -based system in which a plurality of hosts share a large network storage device, COW of Data-block Bit (CDB) to indicate the status of the snapshot, and Data-block Allocation- to indicate the first data block allocated after the snapshot was taken. create / maintain a snapshot state table composed of entries consisting of time Bit (DAB) in memory, A method of performing a snapshot of a mass shared storage device using a memory-based snapshot state table, characterized in that determining whether to perform COW of a data block by checking values of the DAB and CDB after creating a snapshot. The method of claim 1, wherein the snapshot status table, Snapshot creation method of a large shared storage device using a memory-based snapshot status table, characterized in that the snapshot is created and maintained in memory when the snapshot is created, and when there is no other snapshot when the snapshot is deleted. delete The method of claim 1, wherein the COW data block bits (CDB), 1 bit is allocated for each snapshot to allocate the maximum number of snapshots. If '1', COW is executed after snapshot creation. If '0', initial value or status of COW not yet performed is displayed. Snapshot execution method of a large shared storage device using a memory-based snapshot state table, characterized in that. The data block allocation bit (DAB) of claim 1 or 4, Memory-based snapshot state table, which allocates and uses the first bit of every snapshot state table entry and changes the DAB value to '1' when the data block is allocated and used for the first time after the snapshot is created. How to take a snapshot of mass storage device using. A method for deleting snapshots in a storage area network (SAN) based system in which multiple hosts are configured to share a large amount of network storage, A first step of changing the operation mode of all nodes in which the mapping server exists to a snapshot deletion (SNAPSHOT_DESTROY) mode; Obtaining a lock on the mapping block; A third step of processing whether to perform copy-on-write (COW) of the data block indicated by the mapping entry by using the data block allocation bit (DAB) and the COW data block bit (CDB) of the snapshot state table entry; A fourth step of initializing a first allocation bit (DAB) or a COW data block bit (CDB) when the copy-on-write (COW) is performed and unlocking the mapping block; And And a fifth step of deleting the snapshot when all data blocks have been initialized. The method of claim 1, further comprising deleting a snapshot. The method of claim 6, wherein the third step, If the data block allocation bit (DAB) is '1' and the first snapshot, the data block is the first allocated block after the snapshot is created, so the memory-based snapshot status is not deallocated. How to delete snapshot of mass shared storage using table. The method of claim 6, wherein the fourth step If a COW was performed for the current snapshot, the allocated data block is returned and the first allocated bit (DAB) or COW is returned if there is no next snapshot or if copy-on-write (COW) is performed for the next snapshot. A method of performing a snapshot deletion of a mass shared storage device using a memory-based snapshot state table, wherein the mapping block is released after initializing a data block bit (CDB). A method for performing a write operation in a storage area network (SAN) based system in which a plurality of hosts share a large amount of network storage, Determining whether or not a snapshot exists and performing mapping; A second step of reading the mapping block from a disk to obtain a value of a desired mapping entry; Checking a data block allocation bit (DAB) value of a snapshot state table entry to determine whether a data block is first allocated and used after a snapshot is created; A fourth step of allocating a new block, writing the data contents to a copy disk, and changing the datablock allocation bit (DAB) value to 1 when the datablock allocation bit (DAB) value is 0 and the mapping entry value is an initial value. ; A fifth step of determining whether a copy on write (COW) has been performed through a COW data block bit (CDB) value when the data block is allocated before the snapshot; And And a sixth step of performing a copy on write (COW) and changing the COW data block bit (CDB) value to 1 when the COW is not performed as the COW data block bit (CDB) value is 0. How to perform snapshot write of mass shared storage using memory based snapshot status table. The method of claim 9, wherein the sixth step is Reading the snapshot mapping block into a buffer to obtain a snapshot mapping entry and allocating a new physical data block to perform a copy on write (COW); Copying the contents of the data block to the newly allocated data block and recording the copied data block on a disk; And Changing the physical address pointed to by the snapshot mapping entry to the address of the newly allocated data block, changing the CDB value of the snapshot status table entry to "1", and writing the snapshot mapping block to disk. A snapshot write method for a mass shared storage device using a memory based snapshot status table.

Images