CN100504799C - Method for disk snapshot by snapshot pointer - Google Patents

Abstract

The present invention provides a method for taking a disk snapshot using a snapshot pointer. The method is applied in a server. When the server takes a disk snapshot, all disk blocks of the hard disk are planned as a disk volume, and a first part and a second part are planned in the hard disk. A snapshot pointer is established in the kernel space of the memory. The snapshot pointer points to the starting position of the first part. Whenever the original data needs to be modified, it is first copied to the disk block of the second part and used as backup data. The original data is then modified and used as modified data. Then, the comparison information is written into the write synchronous replication comparison table. When the server wants to restore data on the disk volume, the comparison information is read from the write synchronous replication comparison table, the backup data is obtained, and the backup data is overwritten to the corresponding disk block, so that the disk volume can be restored. The present invention can solve the problem of abnormal crash of the operating system caused by exhaustion of kernel space.

Description

How to take a disk snapshot with a snapshot pointer

technical field

The invention relates to a server, in particular to a method for performing a disk snapshot with a snapshot pointer.

Background technique

Currently, general operating systems support a disk snapshot (Snapshot) technology, which can also be called volume shadow copy services (Volume Shadow Copy Services). The so-called disk snapshot is a technology for retaining data in a file system. As the general data backup technology needs to copy the action, the disk snapshot is to quickly archive the disk volume of the hard disk (volume refers to the disk block (chunk) of the hard disk) at any point in time System backup, and when taking a disk snapshot, does not involve any file copying action. And the entire directory structure in the data structure of the file system currently in use is preserved. In this way, real-time creation of disk snapshots of disk volumes at different time points on the hard disk can not only make up for the shortcomings of the traditional backup mechanism, but also, even if the hard disk is damaged or poisoned, the administrator can choose to The disk snapshot replaces the original disk volume (Volume) to minimize the damage.

Based on the above, when the hard disk is taking a disk snapshot, it does not actually copy the file. Instead, when the disk block of the disk volume changes, the original file in the disk block of the disk volume is first copied. The data is sent to other available disk blocks in the hard disk that do not belong to the disk volume, and then the original data in the disk block of the disk volume is modified to become modified data, and a memory of the server is provided inside the operating system (such as: Linux) In the memory space used by the core (kernel) (hereinafter referred to as the kernel space), a copy-on-write table (Copy-On-Write (abbreviation: COW) Table) is established, and the copy-on-write table is written to record each disk area The information in which disk block in the hard disk the original data of the block is written to. The aforementioned technology is called Copy-On-Write (COW) technology.

However, the core memory space of the operating system is limited and will not become larger due to the increase of the system memory. When more disk blocks are changed in the disk volume, the memory space occupied by the core of the operating system when writing the synchronous replication table bigger. When the disk volumes take disk snapshots at different times, or the capacity of the disk volume increases rapidly, the kernel space will be occupied by a large number of write synchronous replication comparison tables, which will exhaust the kernel space and cause the operating system to Exception crashes. Therefore, how to design a method for performing disk snapshots using snapshot pointers to effectively solve the above-mentioned problems is an important issue that needs to be solved urgently.

Contents of the invention

In view of the many problems of the aforementioned existing disk snapshots, the inventor has finally developed and designed a method for disk snapshots using snapshot pointers of the present invention after long-term hard research and experiments. A problem that causes the operating system to crash abnormally.

An object of the present invention is to provide a method for performing disk snapshots with a snapshot pointer, the method is applied in a server, the server includes a memory and a hard disk, wherein the memory is at least planned to provide a kernel space used by the internal core of the operating system , all the disk blocks (chunks) that currently have data on the hard disk are planned as a disk volume. When the server takes a disk snapshot of the disk volume, the first part and the second part are planned in the hard disk, and they are created in the kernel space A snapshot pointer, the snapshot pointer points to the starting position of the first part, and thereafter, whenever the original data in the disk block of the disk volume needs to be modified, the original data is first copied to the second part as backup data, and then the The original data of the disk volume is modified into modified data, and then the comparison information is written in the first part of the write synchronous replication comparison table, and the comparison information includes the address of the backup data and the address of the modified data corresponding to the backup data.

Wherein, the capacity of the first part is equal to the maximum required capacity of the write synchronous replication comparison table; the write synchronous replication comparison table provides for storing a plurality of comparison information; the capacity of the second part is equal to the disk The original data of all the disk blocks of the volume is copied to the hard disk as the required capacity of the backup data.

For the method, wherein, the step of planning the first part and the second part on the hard disk by the server includes:

According to the capacity of the disk volume, it is used to calculate the maximum required capacity of the write synchronous replication comparison table; plan the first part in the remaining space of the hard disk that does not belong to the disk volume ; and planning the second part in the remaining space in the hard disk that does not belong to the disk volume and the first part.

Wherein, the calculation method of the maximum required capacity written into the synchronous replication comparison table includes:

The maximum required capacity written in the synchronous replication lookup table=the number of disk blocks in the disk volume×the capacity of a comparison information; and the number of disk blocks=the capacity of the disk volume/the capacity of disk blocks.

In addition, the position of the first part planned in the hard disk is before the position of the second part in the hard disk; the last part of the first part of the hard disk is directly connected to the last part of the second part. the front part.

When the server modifies the disk block in which the modified data has been written in the disk volume, it directly modifies the disk block of the modified data.

Another object of the present invention is to read the comparison information corresponding to the modified data from the write synchronous replication comparison table first when the server restores the data on the disk volume, obtain the backup data in the second part, and overwrite the backup data You can restore the disk volume to the disk block where it was originally stored.

To sum up, since the address pointer only occupies a very small part of the storage space of the kernel space, and the write synchronous replication comparison table is stored in a hard disk with a storage capacity much larger than the kernel space, the kernel space is no longer written into the synchronous replication comparison Table occupied, can achieve the purpose of saving the storage capacity of the kernel space, and at the same time, when the server takes a disk snapshot, it will not cause the server to be abnormal, so as to achieve the purpose of stabilizing the server.

Description of drawings

Fig. 1 is a schematic diagram of the architecture of the present invention;

Fig. 2 is the action flowchart of the present invention carrying out disk snapshot;

FIG. 3 is a flow chart of the present invention for planning the first part and the second part on the hard disk.

Description of main component symbols:

Server.........5 Storage.............6

Kernel Space......60 Hard Disk......7

Disk Volume...70 Part One...72

Part II ...... 74

Detailed ways

The present invention will be described in detail below in conjunction with the accompanying drawings.

The present invention provides a method for performing disk snapshots with snapshot pointers. Please refer to FIGS. 6. Plan out a kernel space 60 (Kernel Space). When the server 5 performs a disk snapshot on the hard disk 7, it will be processed according to the following steps:

Step 1: All the disk blocks (chunks) that the hard disk 7 currently has data are planned as a disk volume 70 (Volume);

Step 2: plan out the first part 72 and the second part 74 in the hard disk 7, wherein the capacity of the first part 72 is equal to the maximum required capacity of the Copy-On-Write (abbreviation: COW) Table written in synchronous replication, write The synchronous copy comparison table includes a plurality of comparison information, and each comparison information contains the backup data address and the modification data address of the corresponding backup data respectively, and the capacity of the second part 74 is equal to all the disk blocks of the disk volume 70 The original data is copied to the hard disk 7 as the required capacity of the backup data; in an embodiment of the present invention, please refer to shown in FIG. steps, including:

Step 20: First, according to the capacity of the disk volume 70, it is used to calculate the maximum required capacity of the disk volume 70 for writing the synchronous replication comparison table; and in the present invention, the maximum demand for writing the synchronous replication comparison table The calculation formula of capacity is as follows:

The maximum required capacity written in the synchronous replication comparison table = the number of disk blocks in the disk volume 70 × the capacity of a comparison information... (Formula 1)

The number of disk blocks=capacity of disk volume 70/capacity of disk blocks...(Formula 2)

Wherein, "/" in Formula 2 represents a division sign.

For example, assuming that the capacity of the disk volume 70 is 100GB, the capacity of each disk block of the disk volume 70 is 64KB, and the capacity of each comparison information written in the synchronous replication comparison table is 16byte, then write The maximum required capacity for entering the synchronous replication comparison table is (100GB/64KB)*16byte=25MB.

Step 21: plan the first part 72 in the remaining space of the hard disk 7 that does not belong to the disk volume 70; and

Step 22: Planning the second part 74 in the remaining space of the hard disk 7 that does not belong to the disk volume 70 and the first part 72, that is, the end.

In addition, in order to improve the speed at which the backup data is read, in this embodiment, the first part 72 is planned to be at the position of the hard disk 7 before the second part 74 is at the position of the hard disk 7, especially, at the first part of the hard disk 7 The endmost part of 72 is directly connected to the frontmost part of the second part 74, and the backup data can be read at the fastest speed.

Step 3: set up a snapshot pointer in the kernel space 60, and the snapshot pointer points to the initial address of writing the synchronous replication comparison table in the hard disk 7, that is, the initial address of the first part 72 in the hard disk 7; and

Step 4: Whenever the original data in any disk block of the disk volume 70 needs to be modified, the original data is first copied to one of the disk blocks in the second part 74 as backup data, and then the disk volume The original data of booklet 70 is modified into modified data, and the comparison information including the backup data address and the modified data address corresponding to the backup data is written into the write synchronous replication comparison table of the first part 72, that is, the end.

As can be seen from the above, the hard disk 7 stores the write synchronous replication comparison table. Since the storage capacity of the hard disk 7 is much larger than the kernel space 60, the hard disk 7 will not be occupied by the write synchronous replication comparison table. On the other hand, the kernel space 60 only stores the address pointer, and the address pointer usually only occupies a very small part of the storage space, so that the kernel space 60 is no longer occupied by the write synchronous replication comparison table, and the purpose of saving the storage capacity of the kernel space 60 can be achieved. At the same time, when the server 5 performs a disk snapshot, the synchronous replication comparison table is written in the hard disk 7, so that the server 5 is not prone to abnormalities, and the purpose of stabilizing the server 5 is achieved.

In this embodiment, when the server 5 intends to modify the disk blocks that have been written with modified data in the disk volume 70, it directly modifies the disk blocks that have been written with modified data without writing the modified data Stored in the second part 74, so as to keep the backup data of the disk blocks that have been written with modified data in the disk volume 70, so that when the server 5 intends to restore data to the disk volume 70, it will be copied synchronously from writing The comparison table reads the comparison information corresponding to the modified data, obtains the backup data in the second part 74, and overwrites the backup data to the disk block of the modified data, so that the disk volume 70 can be restored.

The above description is only a specific embodiment of the present invention, but the features of the present invention are not limited thereto, and any changes or modifications that can be easily conceived by those skilled in the art within the field of the present invention shall be included in the present invention. In the following patent scope of the present invention.

Claims (7)

1, a kind ofly carry out the method for disk snapshot with the snapshot pointer, be applied in the server, described server includes a storer and a hard disk, it is characterized in that, described storer is cooked up a kernel spacing, when server carries out disk snapshot to hard disk, handles according to the following step:

All disk block that described hard disk has data at present are planned to a disk volume;

Cook up a first and a second portion in the described hard disk, the capacity of described first, be equal to the write the greatest requirements capacity of the synchronization replication table of comparisons, the said write synchronization replication table of comparisons provides a plurality of contrast information that store, the amount of capacity of described second portion, the raw data that then equals all disk block of disk volume is copied in the hard disk as the required amount of capacity of Backup Data;

Set up a snapshot pointer in the described kernel spacing, the start address of the described first of described snapshot pointed in described hard disk; And

When the raw data in any disk block of described disk volume need be modified, earlier described raw data is copied in one of them disk block in the described second portion as a Backup Data, again described raw data is revised and become modification data, and, write writing in the synchronization replication table of comparisons of described first with a contrast information that comprises the modification data address of described Backup Data address and corresponding described Backup Data.

2, method according to claim 1 is characterized in that, described server is cooked up the step of first and second portion at hard disk, comprising:

According to the amount of capacity of described disk volume, in order to calculate the greatest requirements amount of capacity of the said write synchronization replication table of comparisons;

In described hard disk, do not belong in the remaining space of described disk volume and cook up described first; And

In described hard disk, do not belong in the remaining space of described disk volume and described first and cook up described second portion.

3, method according to claim 1 and 2 is characterized in that, the computing method of the greatest requirements amount of capacity of the said write synchronization replication table of comparisons comprise:

Write the amount of capacity of disk block quantity * contrast information of the greatest requirements capacity=disk volume of the synchronization replication table of comparisons; And

The capacity of disk block quantity=disk volume/disk block capacity.

4, method according to claim 2 is characterized in that, the described first position of planning in described hard disk is before the position of described second portion in described hard disk.

5, method according to claim 4 is characterized in that, the part of the described first least significant end of described hard disk, and described second portion part foremost directly continues.

6, method according to claim 1 is characterized in that, when the disk block that has been written into described modification data in to described disk volume when described server is made amendment, directly the disk block of described modification data is made amendment.

7, method according to claim 6, it is characterized in that, when described server desire is carried out the data recovery to described disk volume, read the contrast information of corresponding described modification data from the said write synchronization replication table of comparisons, obtain described Backup Data at described second portion, reach the disk block that described Backup Data is covered described modification data, can recover the disk volume.

Images