No relevant resource is found in the selected language.

This site uses cookies. By continuing to browse the site you are agreeing to our use of cookies. Read our privacy policy>Search

Reminder

To have a better experience, please upgrade your IE browser.

upgrade
Knowledge Base

When making Raid10 on S5300, wrong configuration of mirror disk number lead to bad write capability of the storage

Publication Date:  2012-10-16  |   Views:  220  |   Downloads:  0  |   Author:  SU1000864011  |   Document ID:  EKB1000017266

Contents

Issue Description

Some site uses S5300 storage, in which there are 14 300GB FC disks. The S5300 is linked to a Linux server by FC port. The client demands to do raid10 using the 14 disks and divide 2 1TB LUN, which is mapped to linux server. Then login in the server and test read and write performance of the disks. It costs 5 minutes to create a 1GB file, write performance is bad.

Alarm Information

none

Handling Process

1. Login on the server and uninstall the volume.
2. Login on S5300 and delete lun and raid group.
3. Recreate raid group, raid level is raid10, disk number is 2, there are 14 disks in total.
4. Create 2 1TB luns and map them to the server.
5. Login on the server, rescan the disks, find 2 luns, divide and format them and mount them.
6. Test the performance, it takes 30 seconds to create 5GB file, the performance is increased.

Root Cause

When make raid10, the disk number of mirror group is configured 14, which means 14 disks are in one mirror group, so the capacity of the raid group is as one disk, write performance is bad. Usually we configure disk number as 2, in this case, the capacity of the raid group is as 50% as naked capacity, and the performance is between raid0 and raid5.

Suggestions

On site deployment, we must install and configure according to the directions seriously.