Asymmetrical disk array caching dispatching method

A disk array and cache technology, applied in the field of asymmetric disk array cache scheduling, can solve the problems affecting the I/O performance of front-end applications, unbalanced I/O requests, and reducing the reconstruction speed, so as to shorten the reconstruction time, shorten response time, and improve reliability

CN104503923AActive Publication Date: 2015-04-08HUAZHONG UNIV OF SCI & TECH
4 Cites 8 Cited by

Patent Information

Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Publication Date
2015-04-08

Smart Images

  • Figure 1
    Figure 1
  • Figure 2
    Figure 2
  • Figure 3
    Figure 3
Patent Text Reader

Abstract

The invention discloses an asymmetrical disk array caching dispatching method, belonging to the technical field of computer storage. According to the method, a disk not involved in reconstitution is selected as a sacrificed disk, cache size of the sacrificed disk is reduced, and the cache size of a disk involved in reconstitution is correspondingly increased, so that number of times of input / output (I / O) requests to invalid disks and disks involved in reconstitution are reduced, the reconstitution speed of the invalid disk is improved, I / O response time of users is shortened, the availability and reliability of a storage system are improved, and the performance of the system is improved.
Need to check novelty before this filing date? Find Prior Art

Description

technical field

[0001] The invention belongs to the technical field of computer storage, and more specifically relates to an asymmetric disk array cache scheduling method. Background technique

[0002] Reliability and availability are important criteria for measuring the quality of online storage services. For systems that provide online services for commercial and industrial applications, data loss and often-interrupted service can mean significant financial loss. Redundant Array of Independent Disk (hereinafter referred to as RAID) is one of the mainstream choices that can provide highly reliable and highly available data access services. In the early RAID specifications, such as RAID-1 to RAID-5, it only provides single-disk fault tolerance. In recent years, various RAID specifications with higher fault tolerance have gradually begun to be adopted, because they can still recover all lost data when two or more disks fail at the same time, and at the same time provide uni...

Examples

Embodiment 1

[0041] Taking 1 disk failure in n disks based on the RAID-6 system as an example (n is greater than or equal to 3) below, the asymmetric disk array cache scheduling method of the present invention and the least recently used page replacement algorithm (Least Recently Used, below LRU for short) are combined to describe the specific implementation of the present invention.

[0042] In the LRU algorithm, the weight of the cache page is determined by the storage time interval, which represents the last stored timestamp (timestamp), that is, the access sequence number. In the embodiment of the present invention, the LRU-based asymmetric disk array cache scheduling method needs to maintain the above evaluation method, and which cache page will be rejected is determined by the weight.

[0043] In the embodiment of the present invention, the Buffer Cache structure includes a global LRU linked list and a local LRU linked list, each disk corresponds to a local LRU linked list, the local...

Embodiment 2

[0054] Taking the failure of one of the n disks in the RAID-10 system as an example (n is greater than or equal to 4), the specific implementation of the present invention will be described by combining the asymmetric disk array cache scheduling method of the present invention with the LRU algorithm.

[0055] In the embodiment of the present invention, the Buffer Cache structure includes a global LRU linked list and a local LRU linked list, each disk corresponds to a local LRU linked list in the cache, what the local LRU linked list stores is the cache page of the same disk, and what the global LRU linked list stores is Cache pages for all disks.

[0056] In the case of a disk failure, RAID-10 needs to read the mirror disk of the failed disk to reconstruct the data, that is, the remaining n-2 disks are all sacrificial disks.

[0057] Figure 4 For the flow chart of the LRU-based asymmetric disk array cache scheduling method on RAID-10 of the present invention, specifically co...