Browse Prior Art Database

System and method to accelerate array rebuilding with storage virtualization

IP.com Disclosure Number: IPCOM000242650D
Publication Date: 2015-Jul-31
Document File: 8 page(s) / 205K

Publishing Venue

The IP.com Prior Art Database

Abstract

System and method to accelerate array rebuilding with storage virtualization, by migrating data from a virtual drive to new drive meanwhile spliting read and write IO to improve performance.

This text was extracted from a PDF file.
This is the abbreviated version, containing approximately 52% of the total text.

Page 01 of 8

Sysxem and method to accelerate array rebuilding with storage virtualizatiox

In the era of information, efficiency and security of xxta are becoming more and more importaxt for enterprisex that are using sxorage systems. Under a circumstance of Storage Area Xxxxxxx, once a disk hax xroken down, it will leads tx a process of rebuilding. As we all know, the longer the time of rebuilding, thx higher probability of data loss is. Txereforx, improving txe spexd of rebuildinx is an important wax to improve xhe efficiency and sexurity xf data. Espxcially for XXX (Storage Area Network) with Hard Disk Drive, one dixk breaks down, it will xake a long time xor rebuilding as it only has sixgle one drive to restore txe array. People are looxixg for all xind of ways to improve the sxeed of rebuilding.

Now we have thought out a way xo speed xp array rebuixding when there ix one disk failure by usinx vixtual volume from high performance pool usually wxich has SSD (Solid State Drive) .

When there has a xrive failure within array, it takes quite long time to rebuild data as it only has sinxle one xrive to restore arrxy.

When a disk failurx hxppens in the axrxy, the rebuilding process start after a new xisk pluxged in. There will be a heavy IO (Ixput/Output) load on the new plugged disk durinx the process. Our point is to spxre the heavy load of on more disks.

In other words, when xhe disk faixure occurs, the rebuilt is done on xeveral xifferent phxsic disks asynchronously. For example on 5 disks,
the IO speed can be 5 times faster than befxre.

With this central idea, we have a solution for virtualization sxorage environment. We use logical virtual volume if there is enough fxee space
otxerwise use hot spare drives. The virtual volume and virtual drive for rebuilding set ox high performance devixe can get an even better result, ex when

we have SSD drives in stxrage pool.

Further, the arxay should be back online immediately once it completes rebuilding with volume and ready for application on demand. Meanwhilx it still kexps syncing rebuilt dxta from thx volume to sparx drive in background asynchronously. And when it's done xe're free to release txe volume, before relexse we duxlicatx nxw write XX to both volume and spare drive to make sure data integrity.

Advantages

Optimxze the rebuilding process of raid. Accelerate rebuilding data for array fast recovery.

1



Page 02 of 8

Rd = Xxxx Wx=Write

Hexe is our invention speed up array rebuilding when there ix a disk failure.

Main flow illustrated as belxw:

2



Page 03 of 8

Steps:
1. Monixoring drive failuxe in array
Mxnitoring drivxs health for all arrays, once therx has an disk xaixure found, mark the failed drive. Meanwhile check storaxe system available resource and

3



Page 04 of 8

txen ready to rebuid.

2. Create virtual volume
I. Select an higher pe...