WO2016180049A1 - Storage management method and distributed file system - Google Patents

Abstract

A storage management method and a distributed file system. The storage management method comprises: dividing all the magnetic disks of the distributed file system into a plurality of magnetic disk packets (S101); acquiring the serviceability of each magnetic disk packet, and dividing each magnetic disk packet into a healthy group set, a sub-healthy group set and a bad group set according to the serviceability of each magnetic disk packet (S102); and when data needs to be stored, selecting a magnetic disk packet from the healthy group set, and selecting a disk from the selected magnetic disk packet to store the data (S103).

Description

Storage management method and distributed file system Technical field

This application relates to, but is not limited to, the field of storage applications for distributed file systems.

Background technique

The distributed file system for large-scale application uses the erasure code as the underlying file storage strategy, which can significantly reduce the physical space occupied by files and maintain the same reliability as multiple copies. In enterprise applications, the system scale becomes larger and longer. After the time is running, the disk is damaged or the fault is not a small probability event, but the normal situation that is often faced in daily maintenance.

For example, when an enterprise application generally uses an erasure code, the configuration file storage policy original block N: the check block M is 6:2, and two fragment copies can be damaged at the same time, and the overall reliability of the system is the same as the 3 copy mode. Combined with the current situation of enterprise applications and the inevitable hardware and software problems of the storage cluster itself, the damage of three disks in one storage system may cause partial data corruption. A storage cluster contains hundreds or even thousands of disks, and after a long run, tens of millions of copies are stored on each disk. If the system corrupts more than 2 disks at the same time in a certain period of time, the data content of some files cannot be repaired, resulting in unreadable file contents.

Therefore, how to provide a storage management method that can alleviate the existence of a distributed file system and simultaneously damages more than two disks causes the file content to be unreadable, which is a technical problem to be solved by those skilled in the art.

Summary of the invention

The following is an overview of the topics detailed in this document. This Summary is not intended to limit the scope of the claims.

This paper provides a storage management method and distributed file system to alleviate the existence of distributed file system. At the same time, damage to more than 2 disks will result in unreadable file content.

A storage management method for a distributed file system, comprising: dividing all disks of a distributed file system into multiple disk groups; obtaining serviceability of each disk group according to each disk The serviceability of the packet divides each disk group into a health group set, a sub-health group set, and a bad group set; when data needs to be stored, a disk group is selected from the health group set, and the disk storage data is selected from the selected disk group. .

Optionally, dividing all the disks of the distributed file system into multiple disk groups includes: dividing each disk into multiple disk groups according to the slot information in each disk in the storage server to which the disk belongs.

Optionally, dividing all the disks into multiple disk groups according to the slot information of each disk in the storage server to which the disk belongs includes: dividing the disks having one or more the same slot information in all storage servers into one disk grouping. .

Optionally, obtaining the serviceability of each disk group comprises: obtaining the number of available volumes of each disk group, and determining the serviceability of each disk group according to the storage policy and the number of available volumes of each disk group.

Optionally, the method further includes: when the disk group cannot be selected from the health group set, the disk group is selected from the sub-health group set, and if the disk group cannot be selected from the sub-health group set, the data storage fails.

Optionally, selecting a disk group from the health group set or the sub-health group set comprises: selecting a disk group according to a proportion of available space of each disk group in the health group set or the sub-health group set.

A distributed file system comprising: a plurality of disks, and a location register, the location register comprising: a disk grouping module and a selection memory module, wherein the disk grouping module is configured to: all disks of the distributed file system Divided into multiple disk groups; obtain serviceability of each disk group, and divide each disk group into a health group set, a sub-health group set, and a bad group set according to the serviceability of each disk group; the selected storage The module is set to: select the disk group from the health group set when the data needs to be stored, and select the disk storage data from the selected disk group.

Optionally, the disk grouping module is configured to divide all disks into multiple disk groups according to slot information in each disk in which the disk belongs.

Optionally, the disk grouping module is configured to: divide all disks having one or more same slot information in all storage servers into one disk group.

Optionally, the disk grouping module is configured to: obtain the number of available volumes of each disk group, and determine the serviceable of each disk group according to the storage policy and the number of available volumes of each disk grouping. Sex.

Optionally, the selection storage module is further configured to: select a disk group from the sub-health group set when the disk group cannot be selected from the health group set, and if the disk group cannot be selected from the sub-health group set, the data storage failure.

Optionally, the selecting the storage module is configured to: select a disk group according to a proportion of available space of each disk group in the health group set or the sub-health group set.

A computer readable storage medium storing computer executable instructions for performing the method of any of the above.

The storage management method provided by the embodiment of the present invention firstly manages all disk groups and further distinguishes the serviceability status of the disk group. When storing data, the copy position is preferentially allocated from the health group, and multiple disks (greater than M) appear. In the event of a failure, as long as the disks do not belong to the same disk group, the file contents will not be lost, which can improve the overall reliability of the system.

Other aspects will be apparent upon reading and understanding the drawings and detailed description.

BRIEF abstract

1 is a flowchart of a storage management method according to a first embodiment of the present invention;

2 is a schematic diagram of a distributed file system according to a second embodiment of the present invention;

3 is a flowchart of a step of grouping disks in a storage management method according to a third embodiment of the present invention;

FIG. 4 is a flowchart of a data storage step in a storage management method according to a third embodiment of the present invention.

Embodiments of the invention

The embodiments of the present invention will now be described with reference to the accompanying drawings.

First embodiment:

1 is a flowchart of a storage management method according to a first embodiment of the present invention. In this embodiment, the storage management method provided by the embodiment of the present invention includes the following steps:

S101: Divide all disks of the distributed file system into multiple disk groups;

S102: Acquire serviceability of each disk group, and divide each disk group into a health group set, a sub-health group set, and a bad group set according to serviceability of each disk group;

S103: When the data needs to be stored, select a disk group from the health group set, and select a disk storage data from the selected disk group.

In some embodiments, dividing all disks of the distributed file system into multiple disk groups in the foregoing embodiment includes: dividing all disks into multiple disk groups according to slot information in each storage server to which the disk belongs. .

In some embodiments, dividing all disks into multiple disk groups according to slot information in each storage server of the disk in the foregoing embodiment includes: having one or more same slot information in all storage servers. The disk is divided into a disk group.

In some embodiments, the obtaining the serviceability of each disk group in the above embodiment comprises: obtaining the number of available volumes of each disk group, and determining each disk group according to the storage policy and the number of available disks per disk grouping. Serviceability.

In some embodiments, the above embodiment further includes: selecting a disk group from the sub-health group set when the disk group cannot be selected from the health group set, and failing to select the disk group from the sub-health group set, the data storage fails .

In some embodiments, selecting a disk group from the health group set or the sub-health group set in the above embodiment comprises selecting a disk group according to a ratio of available space of each disk group in the health group set or the sub-health group set.

Second embodiment:

FIG. 2 is a schematic diagram of a distributed file system according to a second embodiment of the present invention. As shown in FIG. 2, in the embodiment, the distributed file system provided by the embodiment of the present invention includes multiple disks 22, and the disks belong to different disks. a storage server, and one or more location registers 21 (Fig. 2 only exemplarily shows a location register), wherein

The location register 21 includes: a disk grouping module 211 and a selection storage module 212,

The disk grouping module 211 is configured to divide all the disks 22 of the distributed file system into a plurality of magnetic disks. Disk grouping; obtaining the serviceability of each disk group, and classifying each disk group into a health group set, a sub-health group set, and a bad group set according to the serviceability of each disk group;

The selection storage module 212 is configured to select a disk group from the health group set when the data needs to be stored, and select the disk storage data from the selected disk group.

In some embodiments, the disk grouping module 211 is configured to divide all disks into a plurality of disk groups based on slot information within each disk in which the disk belongs.

In some embodiments, the disk grouping module 211 is configured to divide a disk having one or more of the same slot information in all storage servers into one disk group.

In some embodiments, the disk grouping module 211 is configured to: obtain the number of available volumes for each disk group, and determine the serviceability of each disk group based on the storage policy and the number of available volumes per disk grouping.

In some embodiments, the selection storage module 212 is further configured to: select a disk group from the sub-health group set when the disk group cannot be selected from the health group set, and if the disk group cannot be selected from the sub-health group set, the data Storage failed.

In some embodiments, the selection storage module 212 is configured to select a disk grouping based on a ratio of available space for each disk grouping in the health group set or the sub-health group set.

The embodiments of the present invention will be further explained in conjunction with application examples.

Third embodiment:

3 is a flowchart of a disk grouping step in a storage management method according to a third embodiment of the present invention. As shown in FIG. 3, the disk grouping step in the storage management method provided by the embodiment of the present invention includes the following steps:

S301: Group all disks at a physical level, and divide all disks into multiple disk groups;

In the distributed file system, the file storage policy uses the erasure code, and the redundancy ratio N:M=6:2 is taken as an example. It is assumed that the distributed file system consists of n storage servers S1, S2, ..., Sn (n ≥ 8). ), and each storage server has d-block disks. For example, a common high-density storage server has 24 or more disks.

The disk of each storage server is divided into multiple disk groups according to the physical slot number s0, s1, ..., sd-1, for example, in the same slot of all storage servers. The disks constitute a disk group, the number of disk groups g is equal to the number of slots d; or a plurality of disks of the same slot form a disk group, and the number of disk groups g is smaller than the number of slots d.

The number of disks and capacity of each disk group can be different. For system read and write performance, you need to ensure that the disk read and write interfaces are balanced.

S302: The distributed file system is powered on, and the location register FLR initializes the data;

When the system is powered on, the FLR initializes the disk packet data area, and all the disks are respectively added to the disk packet data area according to the grouping; when the disk or the node fails, the corresponding disk grouping information is updated. The information of the disk grouping includes approximate real-time statistics such as all disk information, total packet capacity, packet available capacity, packet read/write IO, packet serviceability status, etc., and the data is reported to the file location register FLR for the CHUNK of the file to be allocated. Decision making when copying the location.

S303: FLR calculates a sum of volume weights of each disk group;

The FLR traverses each disk group, counts the number of available disks for all disks grouped by one disk, and calculates the sum of the volume weights, which are included in the disk grouping.

S304: FLR calculates serviceability of each disk group;

The FLR traverses all disk groups and sets the serviceability of the disk group according to certain rules, such as the number of available disks, node failure, network failure, disk abnormality, etc., and adds them to the health group set, the sub-health group set, and the bad group. The set; the group serviceability state refers to whether the disk group can be used by the FLR to allocate the CHUNK copy position, and can include three states: a health group, a sub-health group, and a bad group. The status of the disk grouping is that the number of available disks is greater than or equal to N+M, and the storage unit node fails to maintain reliability, the unit media surface state is normal, and the like; the disk grouping status is sub-health refers to: the disk suddenly drops. Factors such as disk or storage server power loss, resulting in a reduction in the available volume of a packet. When the number of available volumes in a packet/the number of configured volumes in the packet is less than the threshold and greater than or equal to N, the packet is considered to be less serviceable, and the disk grouping status is set to the sub-health group; when a packet available volume is less than N, The disk grouping status is set to bad.

S305: The FLR traverses the disk groups in the above three sets respectively, and maps the disk group according to the number of available volumes.

4 is a flowchart of a data storage step in a storage management method according to a third embodiment of the present invention; As shown in FIG. 4, the data storage step in the storage management method provided by the embodiment of the present invention includes the following steps:

S401: when the client writes the file, the allocated data block copy location message is sent to the FLR;

S402: After receiving the message, the FLR selects the health group set as the current set.

S403: The FLR obtains the random number r, and performs mod (modulo) calculation on the set weight of r and the health group to obtain a keyword key;

S404: The FLR searches for the target disk group by using the binary search in the current set according to the key; if the target disk group is found, step S405 is performed, otherwise step S407 is performed;

S405: In the target disk group, the FLR maintains the reliability condition according to the disk IO equalization, the node failure, the multi-disk failure, and the like, and selects the copy location;

If the number of copies is greater than or equal to N + M (determined by the parameters of the storage policy, generally 6 + 2), then step S406 is performed, otherwise step S407 is performed;

S406: The FLR determines the copy location, writes the data to the corresponding disk, and ends the process.

S407: The FLR selects the sub-health group set as the current set;

S408: The FLR obtains the random number r, and performs mod calculation on the set weights of the r and the sub-health group to obtain a keyword key;

S409: The FLR searches for the target disk group in the current set by the binary search according to the key; if the target disk group is found, step S410 is performed, otherwise step S411 is performed;

S410: In the target disk group, the FLR maintains the reliability condition according to the disk IO balance, the node failure, the multi-disk failure, and the like, and selects the copy location;

If the number of copies is greater than or equal to N + M (determined by the parameters of the storage policy, generally 6 + 2), then step S406 is performed, otherwise step S411 is performed;

S411: FLR fails to allocate a disk, fails to write a file, and fails to exit.

In summary, in the embodiment of the present invention, all disks are first grouped and managed, and the serviceability status of the disk group is further distinguished. When the data is stored, the copy position is preferentially allocated from the health group, and multiple disks (greater than M) appear. In the event of a failure, as long as the disks do not belong to the same disk group, the file contents will not be lost, which can improve the overall reliability of the system.

One of ordinary skill in the art will appreciate that all or a portion of the steps of the above-described embodiments can be implemented using a computer program flow, which can be stored in a computer readable storage medium, such as on a corresponding hardware platform (eg, The system, device, device, device, etc. are executed, and when executed, include one or a combination of the steps of the method embodiments.

Alternatively, all or part of the steps of the above embodiments may also be implemented by using an integrated circuit. These steps may be separately fabricated into individual integrated circuit modules, or multiple modules or steps may be fabricated into a single integrated circuit module. achieve.

The devices/function modules/functional units in the above embodiments may be implemented by a general-purpose computing device, which may be centralized on a single computing device or distributed over a network of multiple computing devices.

When the device/function module/functional unit in the above embodiment is implemented in the form of a software function module and sold or used as a stand-alone product, it can be stored in a computer readable storage medium. The above mentioned computer readable storage medium may be a read only memory, a magnetic disk or an optical disk or the like.

Industrial applicability

In the embodiment of the present invention, all disks are grouped and managed to distinguish the serviceability status of the disk group. When storing data, the copy position is preferentially allocated from the health group. When multiple disks (greater than M) fail, as long as the disks are not If it belongs to the same disk group, it will not cause the file content to be lost, which can improve the overall reliability of the system.

Claims (13)

A storage management method for a distributed file system, comprising: Divide all the disks of the distributed file system into multiple disk groups; obtain the serviceability of each disk group, and divide each disk group into a health group set and a sub-health group according to the serviceability of each disk group Collection and bad group collection; When data needs to be stored, a disk group is selected from the set of health groups, and disk storage data is selected from the selected disk group. The storage management method according to claim 1, wherein dividing the disks of the distributed file system into a plurality of disk groups comprises: dividing all disks into slots according to slot information in each storage server to which the disks belong. Multiple disk groupings. The storage management method according to claim 2, wherein dividing each of the disks into a plurality of disk groups according to slot information in each storage server to which the disk belongs includes: having one or more of the same slots in all the storage servers The disk of bit information is divided into a disk group. The storage management method according to claim 1, wherein the obtaining the serviceability of each disk packet comprises: obtaining a number of available volumes per disk group, and determining each disk according to a storage policy and a number of available volumes per disk grouping Group serviceability. The storage management method according to any one of claims 1 to 4, further comprising: selecting a disk group from the sub-health group set when a disk group cannot be selected from the health group set when data needs to be stored If the disk group cannot be selected from the sub-health group set, the data storage fails. The storage management method according to claim 5, wherein selecting a disk group from the health group set or the sub-health group set comprises: grouping each disk according to the health group set or the sub-health group set The proportion of available space is selected for disk grouping. A distributed file system, comprising: a plurality of disks, and a location register, the location register comprising: a disk grouping module and a selection memory module, wherein The disk grouping module is configured to: divide all disks of the distributed file system into multiple disk groups; obtain serviceability of each disk group, and group each disk according to serviceability of each disk group For health group collections, sub-health group collections, and bad group collections; The selection storage module is configured to: select a disk group from the health group set when the data needs to be stored, and select a disk storage data from the selected disk group. The distributed file system according to claim 7, wherein said disk grouping module is configured to divide all disks into a plurality of disk groups based on slot information in each disk in which the disk belongs. The distributed file system of claim 8, wherein the disk grouping module is configured to divide the disks having one or more of the same slot information in all of the storage servers into one disk group. The distributed file system according to claim 7, wherein said disk grouping module is configured to: obtain a number of available volumes for each disk group, and determine each disk group according to a storage policy and a number of available volumes per disk grouping Serviceability. The distributed file system according to any one of claims 7 to 10, wherein the selection storage module is further configured to: when the disk group cannot be selected from the health group set, from the sub-health group set Selecting a disk grouping, if the disk grouping cannot be selected from the sub-health group set, the data storage fails. The distributed file system of claim 11, wherein the selection storage module is configured to select a disk grouping according to a ratio of available space of each disk group in the health group set or the sub-health group set. A computer readable storage medium storing computer executable instructions for performing the method of any of claims 1-6.

Images