Jeff Baxter

So we've been incredibly proud of the progress that's been going with ONTAP Select.

I don't think anyone expected, and it's certainly not our plan, that

overnight everyone was going to throw out physical hardware and deploy all software to find select right in fact what we've said is that you know our customers are going to continue to use our engineered aff and fast systems for certain workloads then over time adopt on tap select and it's just going to be this great mix and we've certainly over this last year that we've had on tap select available um seen tremendous enthusiasm a lot of people testing it out a lot of people actually deploying it now um and just very rapid growth in it so in 9.2

And in the Select Deploy tool that comes with Select, I believe it's 2.3, we've introduced a lot of new enhancements there.

One of those is the ability to now have two-node HA, which sounds kind of like, wait, you didn't have two-node HA?

That's kind of silly.

But when you don't have shared storage, it actually becomes a little trickier to do two-node HA.

If you have a traditional shared storage system like AFF or FAS or

a lot of our competitors, right?

The actual drives become basically your core and holder, right?

The specialty reservations on the drives let you know which node really has control.

If you're truly in a shared nothing environment, you don't get that luxury, which is why you typically see shared nothing environments, like even like our solid fire, otherwise have a four node minimum, right?

And so we started that way with select where we had a four node minimum as well for HA as well as the single node.

What we're introducing with 9.2 is a way to do two node HA is

primarily for the robo use case, people who just want to put it out in remote offices, right?

And we actually built a mediator directly into the deploy VM.

So that deploy VM that you use to deploy at select now can act as basically a mediator to avoid split brain or loss of quorum scenarios where

Both nodes might be up, but for whatever reason, they've lost contact with each other and neither of them knows which one is primary.

That's a really good recipe for disaster.

So instead, they use a mediator to determine which one has primacy for the workload and can continue functioning even in a split brain scenario.