CN101202758B - Network virtual storage method of multiple clients - Google Patents

Abstract

A network virtual storage method of multiple clients comprises the following steps: loading a public data storage area to a disk of a server, and correspondingly configuring a plurality of special data storage areas; configuring a common cache memory for data storage of the common data storage area and configuring a dedicated cache memory for data storage of each dedicated data storage area; when data writing operation is needed, recording the relevant attributes of the data content to be written, and then sending a data writing instruction and the data to be written to a server; and judging whether the data writing space of the special cache memory is full, and writing the data to be written into the special cache memory according to the judgment result. The method improves the data writing speed of the virtual disk, reduces the disk operation time for reading the data again, and realizes the acceleration of the network virtual storage of the data of a plurality of clients. In addition, the method realizes the optimization of the network virtual storage of the data of the plurality of clients.

Description

Multi-client network virtual storage method

technical field

The present invention relates to a method for network virtual storage, in particular to a method of distinguishing public and private data storage areas in a network virtual storage system and correspondingly increasing public and private high-speed caches (Caches) at the same time, and then storing multiple client data The network virtual storage realizes accelerated and optimized multi-client network virtual storage method.

Background technique

Nowadays, with the continuous improvement of network transmission rate, network storage technology has gained more and more recognition, and NAS (Network Attached Storage, Network Attached Storage) and SAN (Storage Area Network, Storage Area Network) systems are becoming more and more widespread. Land application is the best example. Network storage has many advantages such as cost saving, simple deployment, and convenient operation and maintenance. At present, the network storage technology is developing from the original single data storage to the direction of virtualization, and "virtual storage" has become one of the hottest topics in this technical field. The virtual storage technology transparently presents the storage part to end users, and all its implementation details are completed at the bottom layer. Neither developers nor general usability users need to master the professional knowledge of network storage, let alone know the specific storage components. It is for this reason that network virtual storage is gaining more and more popularity.

Traditional network virtual storage technologies mostly adopt a one-to-one approach, that is, one client corresponds to a certain storage space on the server, and any data read and write operations of the client are completed for its own dedicated storage space. In this way, when the number of clients is large, it will lead to difficult deployment and heavy maintenance workload, especially for the traditional diskless system, it will cause a lot of waste of storage space, and when there are many clients, if a Deploying and maintaining one site will bring a huge workload.

In addition, according to the principle of locality, the data read and write program always accesses a continuous local space within a period of time during the process of performing read and write operations. In this way, multiple clients performing similar data reading and writing tasks will continuously read the same data from the server, thereby increasing the occupation of the system disk. Moreover, because the disk writing speed is generally slow, when many clients simultaneously write data to the virtual disk, it will inevitably cause some operations to wait for a long time, and the average speed of the entire system operation and data storage will be correspondingly slowed down.

Contents of the invention

In order to solve the problems and defects in the above-mentioned traditional technologies, the purpose of the present invention is to provide a multi-client network virtual storage method, which is used to realize the virtual storage system for multiple clients in a local area network virtual storage system with a server and multiple clients. Acceleration and optimization of network virtual storage of data.

A multi-client network virtual storage method provided by the present invention comprises the following steps:

After the system is started, the server loads a public data storage area to the server disk according to the application configuration, and at the same time, dynamically configures multiple dedicated data storage areas on the server disk according to the number of clients; the server is the public data storage area. A public cache memory (Cache) should be configured for the data storage, and a dedicated cache memory (Cache) should be configured for the data storage of each dedicated data storage area; when a client needs to perform a data write operation, the Record the relevant attributes of the data content to be written in the index table, and then send all the data writing instructions and data to be written to the server from the network; and after the server receives the data writing instructions and the data to be written, it will judge the client Whether the data writing space of the dedicated cache memory at the end is full, if not full, then write the data to be written in the dedicated cache memory until all the data to be written are written in the dedicated cache memory; if full, Then, an input/output (I/O) controller on the server replaces a part of the data content in the dedicated cache memory and then writes the data to be written in the dedicated cache memory until it is to be written. All data is written in the dedicated cache memory.

In addition, a multi-client network virtual storage method provided by the present invention also includes the following steps:

After the client's data to be written has been written into the dedicated cache memory, the server's input and output (I/O) controller will coordinate and control the data that has been written into the dedicated cache memory to actually be written into the server's physical disk , and control that the data written in the dedicated cache memory will not be replaced out of the dedicated cache memory within a predetermined time, so as to meet short-term data reading requirements. The client considers the write operation complete when all data has been written to the cache.

In addition, a multi-client network virtual storage method provided by the present invention further includes the following steps:

When the client needs to read data, it first searches in the local index table to determine whether the data to be read is public data. When the relevant attributes of the data content to be read exist in the index table, then The data to be read is private data, otherwise it is public data; the client sends a data reading command to the server; after receiving the data reading command, the server determines the type of data that the client needs to read; when the client needs to read When the data is public data, the server first checks to determine whether the data to be read is stored in the public cache memory, if so, directly reads the data from the public cache memory and returns it to the client, otherwise reads the data from the public data storage area Read the data in and store it in the public cache memory, and then return it to the client; and when the data that the client needs to read is private data, the server first searches to determine whether the data to be read is stored in the private cache memory If yes, read the data directly from the dedicated cache memory and return to the client, otherwise read the data from the dedicated data storage area and store it in the dedicated cache memory, and then return to the client.

In summary, the advantages of the present invention are:

1. In the method of the present invention, multiple clients share the same disk image, so the deployment and switching of the network virtual storage system can be quickly completed.

2. Take the diskless system as an example: Assume that there are 100 clients, and the system data of each client must be stored on the server. If each client system itself needs to use 2.5GB of disk space, 0.5GB is the execution time The amount of data that needs to be overwritten. If adopting the mode of one-to-one correspondence between the client and the disk image in the traditional technology requires: 100*2.5=250GB of disk space, and adopts the method of distinguishing public and dedicated data storage areas in the method of the present invention and requires: 2+100 × 0.5=52GB of disk space, which saves (250-52)÷250=79.2% of the disk space, so the method of the present invention effectively reduces the required data storage space of the client system and saves the entire storage system The hardware disk space and its maintenance cost in .

3. Since the method of the present invention can load a public data storage area onto the server according to the application configuration, and configure a public high-speed cache memory (Cache) correspondingly for the data storage of the public data storage area, it not only satisfies the needs of multiple clients There is a need for parallel reading of public data, and through the public cache memory, the data only needs to be read in once. Since the data reading operations performed by each client are very similar, the hit rate of subsequent data reading will be greatly improved , and then increase the reading speed of the public data to be close to the data reading speed of the internal memory, greatly improving the access and reading speed of the public data by the client.

4. Simultaneously, because the method of the present invention can dynamically correspondingly configure a plurality of special-purpose data storage areas according to the quantity of clients, and respectively configure special-purpose high-speed memory (Cache) correspondingly for the data storage of each special-purpose data storage area, therefore, can The dedicated data read and written by the client is buffered, which reduces the waiting time of a single client, speeds up the reading and writing of data, and reduces the disk operation time and user waiting time for reading out data after writing. Fast writes reduce the average wait time for client write operations for the entire system.

5. Since the method of the present invention adopts the data access mode of same reading and different writing, the data rewriting requests of different clients to the public data storage area will be transferred to respective corresponding dedicated data storage areas, so that the original public data will not be changed all the time. It will be destroyed, avoiding the data error problem that different clients write first and then read in the traditional technology, thereby not only ensuring the safety and reliability of public data but also ensuring that the private data of each client does not interfere with each other.

Description of drawings

Fig. 1 is a system block diagram of a system operated by a multi-client network virtual storage method of the present invention;

Fig. 2 is a method flowchart of a multi-client network virtual storage method of the present invention; and

FIG. 3 is a flow chart of the method when the client performs a data reading operation in a multi-client network virtual storage method according to the present invention.

Explanation of main reference signs:

10 servers

20 clients

30 public data storage area

40 dedicated data storage areas

50 public caches

60 dedicated cache memory

70 index table

Step 100 The server loads a public data storage area to the server according to the application configuration, and simultaneously dynamically configures multiple dedicated data storage areas according to the number of clients

Step 101 configure a public cache memory for the storage of public data, and respectively configure a dedicated cache memory for the storage of private data of each client

Step 102 records the relevant attributes of the data content to be written in the index table

Step 103 sends the data writing instruction and the data to be written to the server

Step 104 judges whether the data write-in space of private cache memory is full

Step 105 replaces part of the data content in the dedicated cache memory

Step 106 writes the data to be written in the dedicated cache memory

Step 107 judges whether the data to be written has all been written into the dedicated cache memory

Step 108 coordinates and controls the written data to be written into the physical disk of the server, and controls the written data not to be replaced outside the dedicated cache memory within a predetermined time

Step 200 searches the index table to determine whether the data to be read is public data

Step 201: the client sends a data read command to the server

Step 202 judges whether the data to be read is public data

Step 203 judges whether the data to be read is stored in the common cache memory

Step 204 judges whether the data to be read is stored in a dedicated cache memory

Step 205 directly reads data from the common cache memory

Step 206: Read data from the common data storage area and store it in the common cache memory

Step 207 directly reads data from the dedicated cache memory

Step 208 reads data from the dedicated data storage area and stores it in the dedicated cache memory

Step 209 returns the read data to the client

Detailed ways

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

Please refer to FIG. 1 and FIG. 2 . FIG. 1 is a system block diagram of a system in which a multi-client network virtual storage method of the present invention operates. Fig. 2 is a method flow chart of a multi-client network virtual storage method of the present invention, as shown in Fig. 1 and Fig. 2 , a multi-client network virtual storage method of the present invention is used to have a server 10 and Accelerating and optimizing the network virtual storage of multiple client data in the area network virtual storage system of multiple clients 20, the method includes the following steps:

After the system starts, in step 100, the server 10 loads a public data storage area 30 to the disk of the server 10 according to the application configuration, and at the same time, dynamically configures a plurality of dedicated data storage areas 30 on the disk of the server 10 according to the number of clients 20 storage area 40;

In step 101, the server 10 configures a public cache memory 50 for the data storage in the public data storage area 30, and configures a dedicated cache memory 60 for the data storage in each dedicated data storage area 40;

When the client 20 needs to perform a data writing operation, at first in step 102, record the relevant attributes of the data content to be written in the local index table 70 for future reading, and then write all the data in step 103 The instruction and the data to be written are sent to the server 10 by the network;

After the server 10 receives the data write instruction and the data to be written, in step 104, it is judged in real time whether the data write space of the dedicated cache memory 60 of the client 20 is full, if not full, then in step 106, the data to be written The data that enters is written in the dedicated high-speed memory 60, then executes step 107; After replacing part of the data content in the dedicated cache memory 60, step 106 is performed again;

In step 107, it is judged whether the data to be written in the client 20 has all been written in the dedicated cache memory 60, if yes, then step 108 is performed, otherwise step 104 is returned, wherein when all the data to be written is written and stored in After being stored in the dedicated cache memory 60, the client considers that the data writing operation has ended, and then can enter into other processing tasks of itself;

After the data to be written in the client 20 has all been written in the dedicated cache memory 60, in step 108, the input/output (I/O) controller of the server 10 will coordinate and control the written data into the dedicated cache memory 60 The data in the server 10 is actually written into the physical disk of the server 10, and the data written in the dedicated cache memory 60 will not be replaced outside the dedicated cache memory 60 within a predetermined time. Therefore, when the corresponding client 20 When the data just written will be read soon, it can be directly read out from the dedicated high-speed cache memory 60, which saves the operation time of reading out from the disk, and speeds up the reading of data to a certain extent. Take the speed.

In the above-mentioned multi-client network virtual storage method of the present invention, since the data writing speed of the cache memory is much higher than the actual physical hard disk, it is equivalent to using a higher writing speed for the client. Hard disk, thereby improving the data writing speed, in addition, when a first client wants to write data, a second client may be writing data to the disk, after adding the cache memory, the first client does not need to communicate with other clients It only needs to write data into its dedicated high-speed cache memory 60. When the data written by each client into the dedicated cache memory can actually be written into the physical disk of the server, the input and output (I/O) controller of the server will carry out subsequent coordination processing. No longer.

Please refer to FIG. 3 now. FIG. 3 is a flow chart of the method when the client performs a data reading operation in the above-mentioned multi-client network virtual storage method of the present invention. As shown in FIG. 1 and FIG. 3 , a method of the present invention A multi-client network virtual storage method may further include the following steps:

When a client 20 needs to perform a data read operation, at first in step 200, a search is performed in the local index table 70 to determine whether the data to be read is public data, when the relevant attribute of the data content to be read exists When it is in the index table 70, it proves that the data to be read has been modified, and is private data, otherwise it is public data;

In step 201, the client 20 sends a data reading instruction to the server 10;

In step 202, after receiving the data reading instruction, the server 10 determines whether the data type to be read by the client 20 is public data;

When the data that the client 20 needs to read is public data, the server 10 first searches and judges whether the data to be read is stored in the public cache memory 50 in step 203, and if so, then in step 205, from the public cache memory 50 Directly read data in, then in step 209, the data read is returned to client 20, if do not store the data to be read in public cache memory 50, then in step 206, read data from public data storage area 30 and stored in the common cache memory 50, and then the read data is returned to the client 20 in step 209, because for the same application requirements, a large amount of same data content is usually read in the same working time period , if it is placed in the public data storage area 30 (i.e. public disk space), it is necessary to repeatedly read the disk, which limits the speed of data acquisition, and adopts the mode of the public cache memory in the method of the present invention , it is only necessary to read the data once, since the data reading operations to be performed by each client 20 are very similar, the hit rate of subsequent data reading will be greatly improved, thereby increasing the reading speed of public data to close to The data reading speed of the internal memory, the larger the public cache memory, the faster the data reading speed;

The step flow for reading private data is similar to that of reading public data. When the data that the client 20 needs to read is private data, the server 10 first searches in step 204 to determine whether the data to be read is stored in a dedicated high-speed In the buffer memory 60, if yes, then in step 207, directly read data from the private cache memory 60, then in step 209, the data read is returned to the client 20, if there is no storage to be read in the private cache memory 60 The fetched data is read from the dedicated data storage area 40 in step 208 and stored in the dedicated cache memory 60, and then the read data is returned to the client 20 in step 209.

In the above-mentioned multi-client network virtual storage method of the present invention, in the process of data reading operation by the client, while accessing and reading the data in the cache memory, it will also be based on the specific access and reading of different data. The situation is recorded, and the input and output (I/O) controller of the server selects the data stored in the cache memory according to the recorded access and read frequency to perform swap-in and swap-out operations to increase the client's direct access to the cache memory. Find and read the hit rate of the required data.

Although the present invention is disclosed above with the aforementioned preferred embodiments, they are not intended to limit the present invention. Those skilled in the art should realize that without departing from the scope and spirit of the present invention disclosed in the scope of the appended claims of the present invention, all changes and modifications are within the scope of patent protection of the present invention. For the protection scope of the present invention, please refer to the appended claims.

Claims (4)

1. A multi-client network virtual storage method, used to realize the acceleration and optimization of the network virtual storage of multiple client data in a local area network virtual storage system with a server and multiple clients, the method comprises the following steps : After the system is started, the server loads a public data storage area to the disk of the server according to the application configuration, and at the same time, dynamically configures multiple dedicated data storage areas on the disk of the server according to the number of clients; The server is correspondingly configured with a public cache memory for the data storage in the public data storage area, and is respectively configured with a dedicated cache memory for the data storage in each of the dedicated data storage areas; When a client needs to perform a data writing operation, it first records the relevant attributes of the data content to be written in a local index table, and then sends all data writing instructions and data to be written to the server through the network; and After the server receives the data writing instruction and the data to be written, it judges in real time whether the data writing space of the dedicated cache memory of the client is full, and if not, writes the data to be written into the dedicated high-speed cache. In the buffer memory until the data to be written is all written in the special cache memory; if it is full, an input and output controller of the server replaces a part of the data content in the special cache memory and then the The data to be written is written into the dedicated cache memory until all the data to be written is written into the dedicated cache memory. 2. The network virtual storage method as claimed in claim 1, further comprising the steps of: After the client's data to be written has been written into the dedicated cache memory, the server's input and output controller will coordinate and control the data that has been written into the dedicated cache memory to actually be written into the server's physical disk , and control that the data written in the dedicated cache memory will not be replaced outside the dedicated cache memory within a predetermined time. 3. The network virtual storage method as claimed in claim 1, further comprising the steps of: When a client needs to perform a data read operation, it first searches in the local index table to determine whether the data to be read is public data. When the relevant attributes of the data content to be read exist in the index When it is in the table, the data to be read is private data, otherwise it is public data; The client sends a data reading instruction to the server; After receiving the data reading instruction, the server determines the type of data that the client needs to read; and When the data that the client needs to read is public data, the server first searches to determine whether the data to be read is stored in the public cache memory, and if so, directly reads the data from the public cache memory. The data is returned to the client, otherwise the data is read from the public data storage area and stored in the public cache, and then returned to the client. 4. The network virtual storage method as claimed in claim 3, further comprising the steps of: When the data that the client needs to read is dedicated data, the server first searches to determine whether the data to be read is stored in the dedicated cache memory, and if so, directly reads the data from the dedicated cache memory. The data is returned to the client, otherwise the data is read from the dedicated data storage area and stored in the dedicated cache, and then returned to the client.

Images