On Monday last week, the VNX Operating Environment code-named “Inyo” was released and made available for download on Powerlink for EMC Partners and EMC Customers. As is customary, there’s a slight delay for “ship from factory” as the default installed software (which happens mid August).
This is a BIG release – with loads of goodies for everyone! I did a preview (including some demos) in the webcast – which is recorded and available here. While I would read the release notes, here are “My Top 10” high notes as far as I’m concerned:
- Mixed RAID in Virtual Pools. This is a material improvement in overall capacity efficiency – about 30-40% for most customers)
- Virtual Pool rebalancing. This is a huge efficiency AND “ease of” improvement. The ability to easily and non-disruptively add IOPs in the form of Flash, or capacity in the form of NL-SAS means that migrations can be avoided, and it’s easier than ever to “start small” and grow as needed.
- Advanced Snaps. I can hear EMC customers and Partners saying “finally”. It’s a fact that our block snapshots in the VNX family needed work. With hundreds of writeable snaps per volume, snaps of snaps, tens of thousands of snaps in total. They still aren’t perfect in my opinion (what is), but we think we’ve made a massive improvement. I’m VERY eager for customer feedback. Ditto on Thin Provisioning overall improvements. Would LOVE to hear customer feedback. We do a lot of testing on this, but inevitably, customers are the truest test.
- VAAI NFS VM-level (aka file) snapshots, but this time with snaps of snaps – and what VMware calls “fast clone” (ergo a deferred snap, not a file-level copy which existed in vSphere 5 and Franklin). This can be leveraged by future vSphere releases to accelerate View Linked Clone use cases and also vCloud Director. There are several folks at VMware and EMC playing with this now – more on performance impact later (I’ve learnt enough from the vSphere 4.1 VAAI experience to wait until we have loads of data before proclaiming victory :-)
- vSphere API for Storage Awareness (VASA) – built in, and with NFS support. We’ve had VASA support from the moment of launch, but it depended on the use of solutions enabler which increased solution complexity and also limited VASA support for block only. With Inyo, VASA is built into the platform – simplifying things – and also supports NAS.
- VAAI XCOPY improvements. I talked about this on this webcast (highly recommended viewing for CX4 and VNX customers) – in Inyo, an optimized code path (“Direct Movement”) is used for many, many more XCOPY scenarioes.
- VAAI Thin Provisioning Reclaim improvements. In future vSphere releases (and also in Windows Server 2012) Thin Provisioning UNMAP is used ever more extensively. These internal optimizations make TP reclaim work better.
- “Flash 1st”. While this is perhaps how FAST VP on VNX should have worked out of the gate – it’s good to optimize. In Inyo – all IOs land on the top tier first, and migrate down as needed.
- FAST Cache improvements. Everything we can do to remove “guardrails” around use cases simplifies things for our customers. Inyo contains some code optimizations
- Some future vSphere optimizations around multipathing (will talk more about this at VMworld at the end of August – hope you’ve registered
So… What’s next? Here’s a hint… The VNX OE (used by the family) releases are codenamed after mountain ranges.
The previous release was “Franklin” – which are 7,192ft/2,192m high and look like this:
The current release is “Inyo” – which are 11,123ft/3,390m high and look like this:
The next release is codenamed “Rockies” – which are 14,440ft/4,400m high and look like this:
So – as big a release as Inyo is relative to Franklin – the Rockies release is much, much bigger :-) More on that later – and also more on Isilon Mavericks which is getting close too….