Start a Conversation

This post is more than 5 years old

Solved!

Go to Solution

468

October 1st, 2009 08:00

Creating Metaluns from different sized raid groups + RAID 1/0 (1+1)

Hello All,

I am now in process of reorganizing disks and raid groups to get better performance on a cx500 flare 19. When I am done, I will have 20 free disks and already have enough spares.

In order to achieve this I am moving production Oracle DB temporarily over to 14 disks. I was thinking of creating 3 raid 10 groups with Metaluns, but puzzled how to split up groups. I could do 3 groups of 2+2 and then a 1+1 by itself for lower IO filesystem (not part of striped metaluns) which is mostly read only binaries

or

I could do 1 group of 3+3 and 2 groups of 4+4 and striped metaluns across those, but I have heard metaluns across different size raid groups is not advisable.

When I am done shuffling around, I will then have 20 free disks to move production back to...

So...questions

1) is it bad to stripe across different size groups
2) would a 1+1 raid 10 group on its own just be bad performance

thnx much!

392 Posts

October 1st, 2009 08:00

1) is it bad to stripe across different size groups

Its not as good as having symmetrical RAID groups in a MetaLUN.

2) would a 1+1 raid 10 group on its own just be bad performance

The answer is wholly dependent on the workload of the LUNs on the RAID group.

The answers to many questions about RAID group performance and MetaLUNs can be found in the EMC CLARiiON Best Practices for Performance and Availability, Flare revision 28.5. A more general explanation of these two can be found in the EMC CLARiiON Storage System Fundamentals for Performance and Availability. Both documents are available on PowerLink.

4.5K Posts

October 1st, 2009 08:00

Answers:

1. yes - it is not recommended to stripe across raid groups with different number of disks
2. 1+1 will provide a level of performance equal to one disk, except if you do all reads

Please see the latest Best Practices guide - it is for release 28, but the method to determine the number of disks based on the host workload needs is still the same. I've included release 26 also as that may have the performance numbers for older disks.

Please see the section on metaLUNs - there are some good suggestions for creating metaLUN - if you get it right, performance will be consistent and scalable.

What you really want to do first is to determine what the workload will be - number of IO/s, IO size, bandwidth requirements, random vs. sequential, etc. You will be better able to create the correctly sized raid groups and LUN once you have a firm understanding of your IO requirements.

EMC CLARiiON Storage System Fundamentals for Performance and Availability

http://powerlink.emc.com/km/live1/en_US/Offering_Technical/White_Paper/H1049_emc_clariion_fibre_channel_storage_fundamentals_ldv.pdf


EMC CLARiiON Best Practices for Fibre Channel Storage: FLARE Release 26 Firmware Update - Best Practices Planning

http://powerlink.emc.com/km/live1/en_US/Offering_Technical/White_Paper/H2358_clariion_best_prac_fibre_chnl_wp_ldv.pdf


EMC CLARiiON Performance and Availability Release 28.5 Firmware Update Applied Best Practices.pdf

http://powerlink.emc.com/km/live1/en_US/Offering_Technical/White_Paper/h5773-clariion-perf-availability-release-28-firmware-wp.pdf

glen

23 Posts

October 1st, 2009 09:00

Thanks for the speedy replies all!

Sounds good...will try to find out the workload
No Events found!

Top