Another part of that announcement was a more formal, 6 monthly release cadence and without disappointment NetApp have this week announced ONTAP 9.1.
Three things really stand out in this release in my opinion;
- New Hardware Platforms – support for more CPU cores, more memory, higher speed Ethernet and fibre channel options.
- Scalable NAS – NetApp FlexGroup – Something this post will dive into later on.
- Flexible Encryption – Volume level encryption for standard SSDs and HDDs.
From those announcements the one I’d like to focus on is the newly announced “Massively scalable containers (FlexGroup)”
What Are FlexGroups?
A FlexVol is a storage construct, a container as such, that is owned by a single node within a disk aggregate, that FlexVol can span the entire aggregate and within reason be grown and shrunk accordingly. However, a FlexVol has a limitation of 100TB in total size (this is platform dependent, so can be smaller on some controllers)
A FlexGroup looks to address these specific challenges.
How does it work? Firstly, a volume is still owned by a node and an aggregate. However, these volumes can be pooled together spanning many nodes within the cluster. This aggregating of volumes addresses both of the challenges of a single FLEXVOL. Firstly reads and writes will then span multiple nodes within the cluster. Secondly the ability to pool together up to 200 FlexVols into a single flexGroup means the single storage container limit rockets from 100TB to 20PB.
Can FlexGroups and FlexVols co-exist? And the answer is absolutely yes. Not all workloads are going to thrive on a FlexGroup so there is still very much a need for Flexible Volumes that we have known and loved for so long.
What about Snapshots? Previously NetApp Snapshots have been created at the FlexVol level so how does this work in a FlexGroup configuration. They are done at the member level meaning they are coordinated across all members. IO is quiesced briefly to ensure consistency across all FlexGroup volumes.
Aren’t these like infinite volumes?
Good question Clustered ONTAP 8.1.1 introduced “infinite volumes”, now I for one did not see any real life scenarios or compelling use cases for infinite volumes, due to a number of limitations that are now addresses in FlexGroups.
In the table below you can see the differences between these infinite volumes and the newly launched FlexGroups.
Who’s going to use FlexGroups?
- Single Namespace
- Simplified Administration
- Integrated Storage Efficiency
- Cluster-Wide metadata scaling eliminates single volume bottlenecks
- Tremendous bandwidth for IO heavy workloads.
These benefits I think can be hugely beneficial with the right use case;
- Large repositories for media content
- Healthcare artefacts (not sure what an artefact is though!)
- Compute intensive NAS workloads
And although infinite volumes had offered the capacity for these workloads, they did not deliver that key element of performance, so well addressed in Flexgroups.
What industries are likely to benefit?
NetApp revamps their entire FAS and AFF lineup! Read about it here: https://t.co/ONuYRdk488 via @ChrisMaki
Tech conferences mean tech announcements – here’s my take https://techstringy.wordpress.com/2016/09/26/shiny-and-new/ via @techstringy
Read about the biggest release in @NetApp’s history with a revamp of their entire hardware range https://t.co/HNh5ok4VKy via @davebrown1969