Hi,
I'm troubleshooting some performance issues on our datastores. I'm curious about our block level storage configuration and whether it is optimal, any advice welcome.
All our disks are 600gb 10k SAS, using an Infortrend storage array. Here's our configuration.....
2 x groups of 7 disks on separate jbod's which make up two raid 6 luns (5+2). These are then combined to present 1 pool of disk at around 5Tb. We have then created 5 x 1 Tb volumes on this pool which are presented as 5 datastores to the hosts.
1) Is there any benefit or drawback to creating 5 separate datastores on the same spindles? Or is it better to have one large datastore.
2) I understand more spindles usually equals more speed but is it better to have smaller groups of disk therefor separating the datastores onto smaller but different groups of physical disk?
I'm worried that essentially all our VM's are using the same backend disk even though they're spread across 5 datastores.
Thanks
