"Tenative target is the next 30 days or less. My feeling is that it will be a day with "teen" in the name,"
Is that a November "teen" or a December "teen"?
Nothing in particular for hyper-v but wodnering how we will be able to upgrade to 2016 with a 3 node cluster. 2016 has a rolling upgrade feature but we would still be missing out on some features like REFS etc as the cluster storage would have already been formatted.
Hoping that Part 2 will have some Hyper-V specific features
We could also move to AHV for our new setup but even that would create a challenge for us namely:
A 3 tier infrastrucure has it downside but I can't see how this would be possible with a 3 nodes nutanix cluster (Actually even with more nodes)
RIght now we have 3 Servers running Hyper-v and on a Power outage, after 30 minutes, we live migrate all VMS from one server and then shut it down to give us more runtime, after another 30 minutes, the same scenario happens until there is only 20-30 minutes left and then all vms are turned off and the last server is then shut down as well. When power is restored, all servers are then powered up and re-balanced automatically without me having to ping a Server or Storage. Doing this with CVM would be suicide i reckon...
RE Asterix Date
I see you caught my dry humor joke. Internally we're targetting the month of November, but this is a really, really big release, so we want to get it right.
RE Hyper-V 2016
I've forwarded this over to our PM for Hyper-V, so he can respond in more detail cc @Raymon
Here's my 2 cents: Keep in mind that Hyper-V 2016 is an absolutely massive new release from Microsoft, which will fundamentally have different considerations for how Nutanix integrates. There are many things to consider; however, the largest is Microsoft's DDA, which is their version of PCI Passthrough.
As you might know, Nutanix uses direct PCI Passthrough of the SAS controller on ESXi, AHV, and XenServer. Hyper-V does not have this same capability in 2012 R2, so we take a different approach currently. We've been clamoring for PCI passthrough for years, to standardize the disk connectivity approach between all four hypervisors we support, and Microsoft is now allowing this in 2016.
That said, Unlike ESXi and AHV, it isn't just a simple passthrough, it requires guest OS kernel changes to support the DDA-enabled devices, so we've had to patch those in, which takes some qualification time.
That's just one low hanging example of what we've got to work through to get 2016 enabled correctly.
RE Hyper-V specific features
Are you looking for something specifically? Or wanting the system to do something it doesn't today? I'm happy to investigate or advocate on your behalf, just let me know.
RE Veeam + AHV
We've talked about this in the past, the ball is in Veeam's court on this one.
RE APC + AHV
APC would need to do work to enhance their PowerChute network shutdown product, that's really the best way to do that. I'll ask our alliances folks to see if we're working with them already.
RE Shutdown sequence
Yes, that would be a challenge with Nutanix (or most SDS products), as in your 3 node cluster, its N+1, so you'd be able to do that with one node, then the whole thing would need to be gracefully shutdown.
I sent him and the PM team an email to get some clarification on this one. I'll bet "yes", but again, this is a massive massive massive shift on Microsoft's part, I have a feeling this is going to be tricky. Looking forward to what Engineering pulls off.
word from product management is that yes, this is the plan to support rolling upgrades from 2012 to 2016. Since the feature isn't done yet, we'll have to wait a bit to see exactly what that looks like
Just adding my name to list of people eager to see this; I'm been holding off implementing our new block as I'd sooner have 2016 "from the begining" rather than upgrading the block at a later date.
Especially if the container has to be formatted as REFS from the get go! I don't think it will be possible to upgrade from NTFS to REFS without destroying the cluster first hence why we are waiting as well.
Now if AHV had VEEAM support we wouldn't be waiting :-p
RE File system
eh, what? The container is not formatted with NTFS now, nor any other windows file system, so we couldn't format it with REFS even if we wanted to.
The container is presented, from a protocol perspective, with SMB3. The actual formatting on the filesystem is invisable to windows.
Much like doing a SMB share from a linux box running a XFS or ZFS file system, or ext or something like that, the protocol and the underlying file system are not coupled together.
You've got that sentence in reverse ... more like if Veeam had AHV support, and let you read into that what you will. Very simply put, we're waiting on them, not the other way around.
It is presented to the cluster as storage, but its SMB3 volume, not a block device that needs to be formatted.
As far as upgrades and such, thats not defined yet, we know we want to support rolling upgrades, but as I mentioned in a prior post, we haven't completed the work on 2016 yet, so we'll have more clarity as time goes on there.