We have a Flash Array from NetApp, which in general does pretty well, but I' trying to decide if mixing workloads on some of the data stores of our more busy VMS is a good idea or not. From the NetApp side we are seeing latency because of CPU processing in some cases, and its because of the workload from my understanding. In the latest case its the archive logs from an Oracle server, I assume its because on a lot of small sequential writes. It was in a shared datastore before and that didn't have the CPU latency like this one is showing now, so I am wondering if it's better to just to move it back. We moved it because compared to the other ones it is substantially larger and is one of our more critical vms.
From my understanding of Netapp arrays they only work on so many volumes at the same time, so in a case like this is worse to have its own data store vs a shared one? We currently use 4.5TB datastores as a standard right now, and I'm wondering if it may be time to increase that or to start doing different size ones from what I've read some people are doing. The size we have has worked out well as it doesn't take too long to evacuate a datastore if we need to, but I've we can get more performance from larger ones I'm wondering if its worth it.
