Unsolved
This post is more than 5 years old
16 Posts
0
1579
February 13th, 2012 10:00
VPLEX - device type (raid) best practice
Anyone have a good white paper on Vplex device types: Raid0, Raid-1, Raid-C and 1:1 mapping ??
I have a Vmax using 3 thin pools of flash (raid5), 15kFC(raid6/14+2) and SATA (raid6/14+2).
I have been using Raid0 for my high IO customers and Raid-C for my lower IO. With the recent vplex code update to 5.01, I can no longer use Raid-C for only one extent, so defaulted to Raid0 for those scenerios instead of using the 1:1 mapping.
Jeff
No Events found!
ankur.patel
2 Intern
•
286 Posts
0
February 13th, 2012 11:00
Jeff what type of info were you looking for? The method you are using makes sense to me since you are using R5/R6 on your backend arrays.
ankur.patel
2 Intern
•
286 Posts
1
February 13th, 2012 12:00
Only reason I can see for doing Raid-C is not having to worry about expanding using the same sized lun when it comes time to do so. If I have a 100GB base and want to expand, I have to get another 100GB lun to keep RAID0, if I need only need 10GB Raid-C would help. Its the same arguments even if VPLEX werent in the picture.
As far as 1:1 mapping I wouldnt look so much at VPLEX best practices as I would at general backend storage best practices. Bad backend storage layout will surface weather you have VPLEX in the mix or not. Most customers are creating large pools of disk and enabling things like FAST VP / FAST / FAST Cache which makes laying out the backend a bit easier. From there they are creating luns and doing a 1:1 map. I havent seen anyone changing things like stripe sizes / cache page sizes ect.
jeffwade_96e6ca
16 Posts
0
February 13th, 2012 12:00
Thanks for the quick reply. Basically wondering why I just wouldn't default to Raid0 for everything. Why would anyone use Raid-C for new allocations? Also, what are the details around 1:1 mapping in regards to performance and best practices.