CN103729150A

CN103729150A - Method for processing Write Hole in memory array

Info

Publication number: CN103729150A
Application number: CN201410016709.0A
Authority: CN
Inventors: 周耀辉
Original assignee: Inspur Electronic Information Industry Co Ltd
Current assignee: Inspur Electronic Information Industry Co Ltd
Priority date: 2014-01-14
Filing date: 2014-01-14
Publication date: 2014-04-16

Abstract

The invention provides a method for processing a Write Hole in a memory array. The processing method comprises the steps that data written by a controller of a home terminal is backed up into a controller of an opposite terminal during RAID 5 degradation in memory systems which form a mutually redundant controller pair, and the controller of the opposite terminal is informed of the releasing of the backed up data after writing is conducted successfully by the controller of the home terminal. If the controller of the home terminal fails when the controller writes on a magnetic disk and the controller of the opposite terminal takes over memory resources of the controller of the home terminal, an RAID 5 is restarted to write the backed up data into a corresponding magnetic disk. Compared with the prior art, the method for processing the Write Hole in the memory array has the advantages that data consistency and data stability of an RAID 5 magnetic disk array are improved, practicality is high, and popularization is easy to achieve.

Description

A kind of method of processing Write Hole in storage array

Technical field

The present invention relates to Computer Storage array technique field, more specifically say the method for processing Write Hole in storage array.

Background technology

RAID5 is " adopting piece interleaving access and check information to be uniformly distributed " pattern.

RAID5 read-write operation is to take band as base unit, and the width of band is 4KB, and the length of band is member's disk number.For the write operation of data in magnetic disk in band, all to upgrade check disk data, to guarantee the consistance of strip data.There is Write Hole problem when reading to rewrite in its write operation.

What is Write Hole? we know in a band of RAID5, must meet D1 ⊕ D2 ⊕ D3 ⊕ D4 ⊕ P=0, and any one data Dx could be calculated by other data and P like this.But the in the situation that of Power Failure, can make D1 ⊕ D2 ⊕ D3 ⊕ D4 ⊕ P unequal to 0.Why? suppose such a case: we will write D1', it is upper that so new P' can write disk, yet unfortunately in the situation that also not writing, just the power supply failed.The situation D1 that at this time may occur and P will become unknown data X and Y.X is likely that D1 is likely also D1' certainly, may be more the potpourri of D1 and D1', and Y is also like this.Data on this band are inconsistent in this case, and if now to have a dish in RAID5 be (the RAID5 degradation) losing efficacy, it is wrong utilizing failure disc that the disk of inefficacy the do not calculate data on this band.Write Hole problem that Here it is.

Based on this, now need a kind of disposal route that can guarantee data consistency on RAID5 band, solve Write Hole problem.

Summary of the invention

Technical assignment of the present invention is to solve the deficiencies in the prior art, and a kind of method of processing Write Hole in storage array is provided.

Technical scheme of the present invention realizes in the following manner, a kind of this method of processing Write Hole in storage array, and its concrete processing procedure is:

The controller storage system of redundancy is each other set, in RAID5 module, increase the data backup operation under degrading state: the RAID5 under degradation writes data and the proof test value that data all will be write at every turn, the band of take is transferred in the redundant manipulator of opposite end as units synchronization, after local controller successfully writes, reinform opposite terminal controller and discharge backed-up data;

If when writing with a brush dipped in Chinese ink disk, controller failure, during redundant manipulator taking over fault controller storage resource, restarts RAID5, before RAID5 provides exterior read-write request, the Backup Data and the verification that do not discharge is write on corresponding disk.

The controller of redundancy is in the situation that RAID5 demotes each other, local controller and opposite terminal controller are Sever/Client pattern, data that local terminal is write are backuped in opposite terminal controller: local controller Client connects with opposite terminal controller Sever, and by local terminal RAID5 information, and the institute band of writing and data send to opposite end backup; Local Data has been write with a brush dipped in Chinese ink, and sends a command to opposite end, allows opposite end discharge backed-up data.

The beneficial effect that the present invention compared with prior art produced is:

A kind of method of processing Write Hole in storage array of the present invention is when local controller is being write with a brush dipped in Chinese ink disk, controller failure, when opposite terminal controller is taken over local controller storage resources, restarting RAID5 writes Backup Data on corresponding disk, and then data consistency, the stability of raising RAID5 disk array, practical, be easy to promote.

Accompanying drawing explanation

Accompanying drawing 1 is implementation structure schematic block diagram of the present invention.

Embodiment

Below in conjunction with accompanying drawing, a kind of method of processing Write Hole in storage array of the present invention is described in detail below.

As shown in Figure 1, now provide a kind of fan design implementation method of effective reduction server energy consumption, its concrete processing procedure is:

The controller storage system of redundancy is each other set, when storage array is paired redundant manipulator, between two redundant manipulators, is designed to high speed interconnected.In RAID5 module, increase the data backup operation under degrading state: the RAID5 under degradation writes data and the proof test value that data all will be write at every turn, the band of take is transferred in the redundant manipulator of opposite end as units synchronization, after local controller successfully writes, reinform opposite terminal controller and discharge backed-up data;

If when writing with a brush dipped in Chinese ink disk, controller failure.During redundant manipulator taking over fault controller storage resource, restart RAID5, before RAID5 provides exterior read-write request, the Backup Data and the verification that do not discharge are write on corresponding disk.So just guarantee the consistance of data on RAID5 band, solved Write Hole problem.

The foregoing is only embodiments of the invention, within the spirit and principles in the present invention all, any modification of doing, be equal to replacement, improvement etc., within all should being included in protection scope of the present invention.

Claims

1. in storage array, process a method of Write Hole, it is characterized in that its concrete processing procedure is:

2. a kind of method of processing Write Hole in storage array according to claim 1, it is characterized in that: the controller of redundancy is in the situation that RAID5 demotes each other, local controller and opposite terminal controller are Sever/Client pattern, data that local terminal is write are backuped in opposite terminal controller: local controller Client connects with opposite terminal controller Sever, and by local terminal RAID5 information, and the institute band of writing and data send to opposite end backup; Local Data has been write with a brush dipped in Chinese ink, and sends a command to opposite end, allows opposite end discharge backed-up data.