We just had a Cx3-20 installed and I talked with them extensively about the first 5 disks and possible uses.
The basic's is that reserve the first 5 disks for the SAN's Microcode. That is the OS that the SAN uses to operate.
You can use the remaining space on these disks to create RaidGroups and LUNs BUT...and this is a big BUT...do not put any high disk IO intensive applications on these disks.
Doing so will affect performance of the SAN. They recommend that if you want to use the space, use it for very low IO activity.
In our situation we are using this space for Quorum drives since cluster quorum disks are written to very little and will not impact performance.
EMC best practices i believe are to not use that set of 5 disks unless for very low I/O apps, even then, they would probably say not to. Its your call obvsiouly. However, in our environment we were able to take advantage of that space as a group of small LUNs for use as snapcache with snapview. Very low I/O and used only when we run snapshots.
Kbing and Mikedeardurff,
Although I agree up to a point, EMC has stated that by adding any LUNs to the first 5 drives will not impact performance 1 bit. This is also noted in the EMC Best Practices Guide for Flare 14, 16, and 19.
You will want to use at least all 5 drives as you will loose 7GB of space per drive. If you decide to use 3 1+1 mirrors, then 6 drives become affected.
Due to the new CX3-20, 40, and 80's which have a much higher capacity for the flare OS, if you were to use all the space on your current array and then wish to upgrade to the newer array, you will need to move the RAID Groups and LUNs off of those drives first.
The most important thing about these first 5 drives is this is where cache is destaged to in case of power failure, controller failure or memory failure. So when cache is flushed out of memory this is where it goes.
This site uses cookies to help personalise content, tailor your experience and to keep you logged in if you register.
By continuing to use this site, you are consenting to our use of cookies.