How to configure new flash pool for SQL specific workloads.

[dougbug]

We are getting new Flash storage in to help alleviate SQL IO latency and performance issues. I'm wondering if there is a special configuration of these disks into an aggregate that will make performance the best it can be for these SQL workloads. Do I just make an aggregate the way i always used too with normal spinning disks or is there some configuration I'd benefit going with that i don't know about.

[Mjizzini · Accepted solution]

Flash Pool is specifically targeted at accelerating repeat random read operations and offloading smallblock random overwrite operations (which are a specific class of writes) from HDDs. Although the SSD cache in a Flash Pool aggregate is a single physical resource within the aggregate, the read cache and write cache are separate logical entities from a cache policy and data management standpoint.
Flash Pool Design and Implementation Guide 
Create a Flash Pool aggregate using SSD storage pool allocation units

[paul_stejskal]

The best way to configure your workloads is to talk to your account team. Depending on the model, SQL can be fairly bandwidth intensive, so you may consider multiple volumes and LUNs to spread work across multiple affinities and SCSI threads (in terms of SAN). The account team can also pull in SQL architects who know how to set this up to best perform and run good.