AFF and StorageGrid best practice

[FelixZhou]

We have a six node AFF cluster, it is currently tight on space usage, is that a good idea to expand it with StorageGrid or add a HA pair of FAS? we have mixed  data of CIFS, LUNs for SQL, Exchange, ESX and other applications. we know we have some old image data in CIFS servers, may be good for StorageGrid, we only keep snapshot a few weeks locally, there are no much we can save to move snapshots to StorageGrid, can we use StorageGrid for high performance required volumes such as SQL, ESX or other application. Worry about performance issues. Can someone share some experience about successful examples about using StorageGrid?

[ttran]

Hello @FelixZhou ,

It would be best to add more shelves directly to the AFF/FAS or full HA pairs to the Cluster than using StorageGrid in a FabricPool implementation for high-performance demanding workloads. Mainly because you would be using the local AFF/FAS storage for the initial staging until the data is moved to StorageGrid, which is mainly used for "cold" or less in-demand data. Your account team can give you a better suggestion after they size your workload demands. StorageGrid and FabricPool have its niche usage, but high workload is what an AFF is designed for whereas StorageGrid isn't even on the same playing field when looking at overall performance and/or full feature sets.

Regards,

Team NetApp

[paul_stejskal]

So snapshot only FabricPool tiering isn't a bad way to go. Honestly I would talk to your account team. I know for a certain customer we ended up moving them to snapshot only on several critical volumes with high change rate and it was quite brilliant really.