ORIGINALLY PUBLISHED 10th May 2017

Hi everyone,

The last couple of days have been the UK user group again, and these interesting discussions always give me a number of new topics to talk about on the blog.  The best quote from the User Group can be attributed to @barrywhyteibm

“No matter what the question – DRAID 6 is the answer”.

Before I go in to what we mean by that – if you are interested in coming to any of the SVC/Storwize User groups – then the development team try to keep a list of all the known user groups here:

http://www-01.ibm.com/support/docview.wss?uid=ssg1S1009304

So – why is DRAID 6 the answer….

The 7.7.1 DRAID 6 solution has been tested to show that the IO performance of DRAID 6 is faster than Traditional RAID 5 in almost all workloads.  There are likely to be two things about that statement that may make you doubt me:

1/ I’ve heard that RAID 6 is slower than RAID 5.  Well I wrote a whole blog post about that very topic, which I’ve linked to below.  Hopefully this will help you to understand why this doesn’t affect storage controllers and real-world workloads:

https://www.ibm.com/developerworks/community/blogs/svcstorwize/entry/RAID_5_Versus_RAID_6?lang=en

2/ OK… That’s easy to say – prove it.  Well here is a direct comparison of Traditional RAID 5 versus Distributed RAID 6.  This test was run with the same physical drives on the same machine in our lab.  The R5 test is using 5 Arrays of 9 drives each in a single pool, and the DR6 test is using a single DRAID 6 array. 

And one final thing – we all know that Data is now the life-blood of our businesses.  There is almost no one out there that can afford to lose data.  When even the “small” 15k Drives are reaching 1TB in size, the rebuild times are getting so high that the potential for a second drive failure during the rebuild phase is nowhere near as low as it used to be.  A 1.2TB 10k drive will take 4 times longer to rebuild than a 300GB 10K drive did.  But those bigger drives are no more reliable than the smaller ones were (they are no worse either).  So now the chance of a double drive failure is 4 times bigger than it was with a 300 GB drive.

In Summary – you can’t afford to use RAID 5 any more.  And if you are going to use RAID 6 – why not use DRAID 6 for all the performance enhancements and benefits.

If you want some advice about how to configure DRAID – especially for use behind an SVC, then Barry’s done a really good blog post on this recently:

https://www.ibm.com/developerworks/community/blogs/storagevirtualization/entry/Configuring_RAID_DRAID_Best_Practice_SVC_and_Storwize?lang=en

One final point – There are some important DRAID fixes included in 7.7.1.5 and 7.7.1.6, so ideally you should get onto those code levels or higher for using DRAID.

Hope this has been helpful

Andrew