We are looking at using FlexGroup in a single region since we need more throughput that a single HA pair can offer. Striping a volume across multiple HA pairs would give us that much needed cumulative performance. To restrict costs, we would like to have a backup (manual failover) in a different region and hence looking at a single-AZ FlexVol as a SnapMirror target.
#Is it possible to SnapMirror a FlexGroup to a FlexVolume (cost effective DR solution)?
warm shale
bold nacelle
Hey, sorry, but no, that is not supported even on prem.
warm shale
I sort of guessed but was hoping against hope. IIRC, snapmirror is block level and requires a very identical target topology.
I worked at NetApp developing Infinite volume and FWIK, FlexGroup took the consistent SnapMirror code/logic from Infinite volume.
frigid finch
How would that save costs? Just wondering. You can have a FlexGroup on a single node.
warm shale
I am not sure if there are minimum requirements for FlexGroup target compared to a FlexVol.
frigid finch
It can't be converted either way, but even if it could, the cost would be the same because it's based upon total size of aggregate.
(disks used specifically)
cunning oak
for FSxN, customers pay for the throughput capacity of their file system as well as the amount of SSD that has been provisioned
you would pay the same for a 1TiB file system that has a single 100GiB FlexVol versus a 1TiB file system with an 8-constituent 100GiB FlexGroup
warm shale
We have a single FlexVol in a dedicated FSxN and this has become a bottleneck for metadata intensive operations.
We would like to explore FlexGroup to spread the metadata load across multiple nodes. We want to scaleout for performance and not capacity.
Since we do not require additional storage capacity but more compute and distribution of load, how would that impact the cost?