KR20050114028A - Cooperative caching device and method for san file systems - Google Patents

Abstract

Disclosed are a cooperative cache apparatus and method for a SAN file system. The location manager manages location information of allocated blocks. When a page failure manager fails to find a block in the local cache area, the page failure manager sends a query to another system to send the block to the system corresponding to the location information of the block provided from the received system. request. The transmission manager transmits the requested block to other systems that requested the block. The operating system buffer manager checks the validity of the block retrieved by the location manager, terminates the processing if the block is valid, and requests the page fault manager to process the page fault. According to the present invention, the cache information of the systems distributed in the SAN environment are shared with each other to form a virtual larger cache, thereby increasing the cache hit rate, reducing the number of disk accesses, reducing the load on the server, and improving the performance of the cache system. Can be improved.

Description

Cooperative Caching device and method for SAN file systems

The present invention relates to a cooperative cache management apparatus and method used in a SAN file system environment, and more particularly, to cache data of each system in order to reduce the number of disk accesses and to reduce the cost of transmitting and receiving messages between systems in a SAN file system environment. A cooperative cache apparatus and method for sharing, managing, and using the same.

1 is a diagram illustrating an operation structure of a conventional cooperative cache system.

Referring to FIG. 1, the existing cooperative cache algorithms used in a storage area network (SAN) file system environment exchange virtually the cache information managed by each of the distributed systems through mutual cooperation. virtual) form a large cache. As the size of the cache increases, the global cache hit ratio increases, and the number of times the server's disk is accessed to find data that does not exist in the cache is reduced. This reduces the load on the server and speeds up data access time compared to using a single cache system that uses only a small local cache.

In the conventional cooperative cache algorithm, a single server manages all shared cache information and a plurality of servers divide and manage them. When one server manages the cache information, the server managing the cache information is overloaded, which slows down the information retrieval time of the global cache, thereby increasing the data access time. When multiple servers divide and manage cache information, the problem of overloading one server does not occur, but the number of messages exchanged for managing and using cache information between servers managing divided cache information increases. As the number of messages increases, a communication cost used for transmitting and receiving messages increases, and a load of a server for processing a message increases.

The technical problem to be achieved by the present invention is to share and manage the cache information of multiple systems in the SAN file system environment, can process a new data structure that can reduce the number of messages sent and received for managing cache information, the number of messages The present invention provides a cooperative cache management apparatus and method used in a SAN file system environment that can maintain consistency while reducing the complexity.

Another technical problem to be achieved by the present invention is to share and manage cache information of various systems in a SAN file system environment, and to process a new format data structure that can reduce the number of messages sent and received for managing cache information, The present invention provides a computer-readable recording medium that records a program for executing a cooperative cache management method used in a SAN file system environment that can be used in a SAN file system environment to reduce the number and maintain consistency.

In order to achieve the above technical problem, a cooperative cache device for a SAN file system according to the present invention, the location manager for managing the location information of the allocated blocks; When a search fails in the local cache area, a query for searching for a location of a desired block is transmitted to another system, and a request for transmission of a block is made to a system corresponding to location information of a block provided from the system receiving the query. Page failure manager; A transmission manager for transmitting the requested block to another system requesting the transmission of the block; And an operating system buffer manager for checking the validity of the block retrieved by the location manager, terminating the processing if the block is valid, and requesting the page failure manager to process the page fault.

In order to achieve the above technical problem, the cooperative caching method for a SAN file system according to the present invention, each system shares and manages the cache information in an environment where a plurality of systems share a storage device attached to the SAN A cooperative cache method, comprising: searching for a block in a local cache area of a first system by a first system; If the first system fails to retrieve the block, forwarding a query to a second system of the plurality of systems to search for a location of a desired block; Requesting, by the first system, the block to be transmitted to a system corresponding to the location information of the block received from the second system; And receiving, by the first system, the block from the system that requested the transmission of the block.

As a result, the cache information of the systems distributed in the SAN environment are shared with each other to form a virtually larger cache, thereby increasing the cache hit rate, reducing the number of disk accesses, reducing the load on the server, and improving the performance of the cache system. Can be.

Hereinafter, with reference to the accompanying drawings will be described in detail a preferred embodiment of the cooperative cache apparatus and method for a SAN file system according to the present invention.

2 is a diagram illustrating a configuration of a cooperative cache device for a SAN file system according to the present invention.

Referring to FIG. 2, several systems 210, 220, and 230 constituting a cooperative cache device for a SAN file system according to the present invention share storage devices 250 to 280 attached to the SAN 240. The SAN 240 directly connects the storage devices 250 to 280, which are individually connected to servers (not shown), to a high-speed dedicated network such as an optical channel to enable centralized management of the storage system. In addition, the SAN 240 accesses and stores data stored in the storage devices 250 to 280 more effectively in an environment of directly accessing the storage devices 250 to 280 connected to the network without going through a server.

FIG. 3 is a block diagram showing a detailed configuration of each system named Node A 210, Node B 220, and Node C 230 shown in FIG.

Referring to FIG. 3, each system 210-230 includes a CLM 310, an LM 320, a CM 330, an LLM 340, a PFH 350, a DM 360, an AM 370, And a buffer 380.

The cluster lock manager (CLM) 310 manages locks of the entire system, and the local lock manager (LLM) 340 manages its locks.

The location manager (LM) 320 is a location information manager located in each system 210 to 230 to manage location information of blocks assigned to the system, and search for block location to provide a search result. Each system 210 to 230 queries the LM 320 to retrieve the block position at the time of a block request or forcibly invalidating the block information, at the latest block request, or at a duplicate check of the block. If the block search is successfully completed in the local cache area, it is said to be a local hit. If the block search fails in the local cache area, a page fault is generated.

Consistency Manager (CM) 330 is a manager that performs the task of maintaining the consistency of the cache. In the present invention, a Lazy Write Invalidate (LWI) algorithm, which is a method of changing a block when the changed block is accessed without changing the content immediately, is used.

The Page Fault Handler (PFH) 350 performs all procedures to solve the problem when a page fault occurs, that is, when the system 210 to 230 fail to retrieve the desired block from the local cache area. I'm an administrator. Accordingly, the PFH 350 inquires the LM of another system to search for the position of the desired block at the page fault and requests the DM of the system corresponding to the position information received from the LM of the system to transmit the desired block. .

The delivery manager (DM) 360 transmits the block to the systems 210 to 230 that have requested the block of the system.

Age Manager (AM) 370 is an administrator who provides information for selecting the most inactive systems 210-230. The present invention sends the block that will find and replace the most inactive system (i.e., a system with a large number of blocks that have not been referenced for a long time) at cache replacement, and the AM 370 manages the information about it.

An operating system buffer manager (OSBM) 380 checks whether a retrieved block is valid and performs an operation according to whether the block is valid. The OSBM 380 terminates the processing if the block is valid, and requests the page fault processing from the PFH 350 if the block is invalid. In addition, the OSBM 380 verifies the validity of the information in order to reduce the number of disk accesses and to check whether there is duplicate information in the cache of another system.

A diagram illustrating a data structure for managing meta information describing block information in a cache in a cooperative cache apparatus according to the present invention.

Referring to FIG. 4, the meta information includes information about a block's current position, whether a block is changed, whether a block is overlapped, and whether a block is newest. The meta information is managed in units of files by the LM 320, and the LM 320 manages information of blocks that each file is in charge of using a location table. In addition, all files are distributed and managed in the respective systems 210 to 230 by a round robin method. The structure of the location table consists of an inode number created to store information about each file, a block number according to the inode, and location information of blocks. Conventional techniques can reduce the use of information storage space by recording only the information in units of bits in the present invention, compared to recording information in units of bytes to store and manage such information.

The location information of the blocks is allocated 2 bits for each system 210 to 230 so that the first bit indicates whether there is a corresponding block, and the second bit indicates whether the corresponding block is up to date. Referring to the table shown in FIG. 4, it can be seen that the system owning block 19 of file number 2 is a system N1 and a system N3 as the system in which the first bit of the system is set to 1 for block 19. It can be seen that the system N1 having the second bit set to 1 has a block of the latest information. In the cooperative cache device for the SAN file system according to the present invention, each system 210 to 230 first searches its local cache area to access a desired block. When a page fault occurs, it searches the cache area or disk of another system to access the desired block.

FIG. 5 is a diagram illustrating a process of finding cache information in a local system constituting a cooperative cache apparatus according to the present invention.

Referring to FIG. 5, the OSBM 380 transmits a validation query to the CM 330 to confirm whether a retrieved block is valid (①). The CM 330 verifies the validity of the block and notifies the OSBM 380 whether it is valid (2). The OSBM 380 terminates the processing if the block is valid, and requests the page fault processing from the PFH 350 if the block is invalid (③). The PFH 350 inquires the LM 320 of the latest information position of the block (④), and receives a response from the LM 320 (⑤). The PFH 350 transmits a block request to the system having the latest information (6), and the DM of the system receiving the block request transmits the requested block to the PFH 350 (7). The PFH 350 transfers the received block to the OSBM 380 and terminates the processing (⑧).

FIG. 6 is a diagram illustrating a processing process when cache information is not found in a local system constituting a cooperative cache apparatus according to the present invention.

Referring to FIG. 6, if the system 210 to 230 fails to retrieve the desired block in the local cache area (ie, a page fault has occurred), the OSBM 380 may retrieve the PFH 350 to retrieve the block's location. ) To request processing (①). The PFH 350 transmits a query to the LM 320 (2), and the LM 320 retrieves the block position and responds to the PFH 350 (3) in response thereto. If the block does not exist in the cache area of another system, the PFH 350 ends the page fault processing by instructing the block to be transmitted from the disks 250 to 280. If the block exists in the cache area of another system, the PFH 350 requests the DM of the system having the block to transmit the copy of the block (④). The DM of the system transmits a copy of the block to the PFH 350 (⑤). The PFH 350 completes the page fault process by delivering the received block to the OSBM 380 (6).

In the cooperative cache apparatus for the SAN file system according to the present invention, unlike the prior art, the location information of the block changed by the operation of writing and reading is changed without additional message exchange. As described with reference to FIG. 6, when a page fault occurs, a query is made to the LM managing the block to determine the location information of the block. Since this query means bringing a copy of the block to the requesting system, the requested system part of the block location information is duplicated at the same time as the search. If the block transmission fails, the DM in charge of the block transmission notifies the LM of the transmission failure, readjusts the location information, and completes the work.

FIG. 7 is a diagram illustrating an example of a location information table data structure before the system N1 constituting the cooperative cache device according to the present invention performs a block read operation and after a read operation.

Referring to FIG. 7, when the system N1 wants to search for a block with block ID 24 for a file having an inode number 2, first acquires a lock on the file and then searches for a location of a block for which a read operation is requested in the location information table. . At the same time as the search, the block status is recorded as having the block requested by the system. That is, the first bit indicating whether or not the block exists and the second bit indicating whether the block is up to date are all changed to 1.

8 is an example of a location table data structure before and after a write operation is performed on a block whose block ID is 24 by a system N3 constituting a cooperative cache device according to the present invention; Figure is a diagram.

Referring to FIG. 8, the system N3 first acquires a write lock on a file, and then searches the block position information of a block to be written in the position information table in the same manner as a read operation. Only the system requesting the write operation has a block of the latest information, which changes the location information of the block. In the prior art, all of the systems having blocks that are not up-to-date information are sent a block removal message to remove the block from the cache of each system. However, if this process is performed every time, the number of messages increases and the system load increases, resulting in a long processing time and a decrease in performance. Therefore, the cooperative cache apparatus according to the present invention uses a method for delaying such message exchange as much as possible and changes only the location information of systems having existing information to be removed, thereby effectively reducing the load caused by the message exchange of the system. That is, the first bit indicating whether the corresponding block of the system requesting the write operation is owned and the second bit indicating whether it is newest are replaced with 1, and the location information of the remaining systems that own the previous block is up to date. Change only the second bit of all to 0.

9 is a diagram illustrating a process of replacing cache information in the cooperative cache apparatus according to the present invention.

Referring to FIG. 9, the OSBM 380 first checks the validity of the validity of the CM 330 in order to reduce the number of disk accesses and check whether there is duplicate information in the cache of another system. Perform the processing step (①). The CM 330 queries the LM 320 whether the corresponding block is valid (②). The LM 320 determines whether the requested block has been changed or not, and sends a response to the CM 330 that it is "valid" if it is not changed by the other system or is the latest information. (③). If the block is not valid, the CM 330 responds to the OSBM 380 to immediately delete the replacement block and terminates the process (④). On the other hand, when another system having the latest information and the corresponding block is duplicated, the DM 330 responds to the OSBM 380 to immediately delete the replacement block and terminates the process (④). . On the other hand, if the corresponding block is the latest information and there is no other system having the latest information overlapping, and if there is a system having the previous information, the LM 320 does not have the latest information in step 3). Returns an Invalidating List consisting of a list of. The CM 330 forcibly transmits the invalidation message for the block to the systems included in the invalidation list and completes the replacement process for the cache information (⑤).

The cooperative cache management apparatus and method used in a SAN file system environment according to the present invention employs a new method different from the methods used in the prior art for selecting cache information to be replaced in the process of replacing the aforementioned cache information. In addition, the apparatus and method for cooperative cache management according to the present invention searches for an inactive system, that is, a system having a large amount of cache information that has not been referenced for a long time, and replaces the cache information that has not been used for the longest time in the system. Select. At this time, when selecting an inactive system, a policy is used to select a system which is generally predicted to be inactive instead of accurately selecting the most inactive system. Choosing the most inactive system with the highest accuracy may yield better performance with the best policy, but it can slow down performance due to the increased management and message communication costs required to maintain information about the inactive system. . Therefore, considering the cost of maintaining information, a policy is used to select a system that is generally inactive at a cost that does not significantly affect performance.

On the other hand, the cooperative cache management apparatus and method according to the present invention includes an age information manager to perform the above-described policy. Each system divides the elapsed time before a block is first used and then used again by N steps, meaning that the higher the step, the longer the elapsed time. The number of blocks corresponding to the highest level is maintained, and this is defined as age information of each system. Each system exchanges age information with other systems through the Age Information Manager. The Age Information Manager generally manages the age process to select inactive systems by: In order to reduce the number of additional message exchanges used for the exchange of age information between systems, the number of additional messages is first reduced by attaching the current age information of each system as a block search or a block transmission between systems. And for a system that has not exchanged messages with each other for a certain time, the age information is collected by forcibly exchanging messages to update the age information with the latest information. Therefore, the number of messages used for age information exchange is minimized, and age information management between systems is performed. The age information manager stores the age information of each system as an age list, and uses this information when selecting the replacement target cache information to consider a system having a lot of high age information as an inactive system. Process.

The invention can also be embodied as computer readable code on a computer readable recording medium. The computer-readable recording medium includes all kinds of recording devices in which data that can be read by a computer system is stored. Examples of computer-readable recording media include ROM, RAM, CD-ROM, magnetic tape, floppy disk, optical data storage, and the like, and may also be implemented in the form of a carrier wave (for example, transmission over the Internet). Include. The computer readable recording medium can also be distributed over network coupled computer systems so that the computer readable code is stored and executed in a distributed fashion.

Although the preferred embodiments of the present invention have been shown and described above, the present invention is not limited to the specific preferred embodiments described above, and the present invention belongs to the present invention without departing from the gist of the present invention as claimed in the claims. Various modifications can be made by those skilled in the art, and such changes are within the scope of the claims.

According to the cooperative cache apparatus and method for a SAN file system according to the present invention, a new format data structure for sharing and managing cache information of multiple systems in a SAN file system environment is developed to reduce the storage space cost required for managing cache information. You can. In addition, by reducing the number of messages sent and received for managing cache information, it is possible to improve performance and reduce communication costs by reducing the load of each system in the SAN file system. Furthermore, by effectively reducing the number of messages exchanged between systems in order to maintain the consistency of cache information and to select caches to be replaced, the performance and performance of the system can be improved by reducing the communication cost and load between systems in the SAN file system. do.

1 is a view illustrating an operation structure of a conventional cooperative cache system;

2 is a diagram illustrating a configuration of a cooperative cache device for a SAN file system according to the present invention;

FIG. 3 is a block diagram showing a detailed configuration of each system named Node A 210, Node B 220 and Node C 230 shown in FIG.

4 is a diagram illustrating a data structure for managing meta information describing block information in a cache in a cooperative cache apparatus according to the present invention;

5 is a diagram illustrating a process of finding cache information in a local system constituting a cooperative cache apparatus according to the present invention;

FIG. 6 is a diagram illustrating a processing process when cache information is not found in a local system constituting a cooperative cache apparatus according to the present invention; FIG.

FIG. 7 is a diagram illustrating a process of changing a data structure when a system constituting a cooperative cache device requests to read cache information according to the present invention; FIG.

FIG. 8 is a diagram illustrating a process of changing a data structure when a system constituting a cooperative cache device requests to write cache information according to the present invention. FIG.

9 is a diagram illustrating a process of replacing cache information in the cooperative cache apparatus according to the present invention.

Claims (9)

A location manager for managing location information of allocated blocks; When a search fails in the local cache area, a query for searching for a location of a desired block is transmitted to another system, and a request for transmission of a block is made to a system corresponding to location information of a block provided from the system receiving the query. Page failure manager; A transmission manager for transmitting the requested block to another system requesting the transmission of the block; And And an operating system buffer manager which checks whether the block retrieved by the location manager is valid, terminates the processing if the block is valid, and requests page fault processing from the page failure manager if the block is invalid. Cooperative cache device for SAN filesystems. The method of claim 1, Cooperative cache device for the SAN file system, characterized in that it further comprises a validator for maintaining the consistency of the cache, and changes the block when accessing the block requiring a change. The method of claim 1, And a aging information manager for managing information required to transmit a block that will find and replace a system having a longest number of unreferenced blocks at the time of cache replacement. The method according to any one of claims 1 to 3, The location information of the block is a cooperative cache device for a SAN file system, characterized in that it comprises a first bit indicating whether the block exists or the second bit indicating whether the latest information of the block. A cooperative cache method in which each system shares, manages, and uses cache information in an environment in which a plurality of systems share a storage device attached to a SAN. A first system searching for a block in its local cache area among the plurality of systems; If the first system fails to retrieve the block, forwarding a query to a second system of the plurality of systems to search for a location of a desired block; Requesting, by the first system, the block to be transmitted to a system corresponding to the location information of the block received from the second system; And And receiving, by the first system, the block from the system that requested the transmission of the block. The method of claim 5, Checking whether a change to the block is required; And If it is determined that a change is required, changing the block when the corresponding block is accessed. The method of claim 5, Dividing the age information into predetermined steps according to the usage time of the cache information; Extracting and storing only the number of the highest steps among the cache information divided into the predetermined steps; And Exchanging the age information of each system at the time of searching and exchanging cache information without a separate message exchange, and exchanging additional age information message only in a system in which cache information retrieval and exchange have not occurred for a predetermined period of time. Cooperative cache method for SAN filesystems. The method according to any one of claims 5 to 7, The location information of the block is composed of a first bit indicating whether the block exists, and a second bit indicating whether the latest information of the block. A computer-readable recording medium having recorded thereon a program for executing a cooperative caching method in which each system shares, manages, and uses cache information in an environment in which a plurality of systems share a storage device attached to a SAN, A first system searching for a block in its local cache area among the plurality of systems; If the first system fails to retrieve the block, forwarding a query to a second system of the plurality of systems to search for a location of a desired block; Requesting, by the first system, the block to be transmitted to a system corresponding to the location information of the block received from the second system; And And receiving, by the first system, the block from the system that requested the transmission of the block. A computer-readable recording recording a program for executing a cooperative cache method for a SAN file system on a computer. media.