VVols: A Whole New World for SQL Server Virtualization

Ah, VVols. The VMFS Datastore killer. And very soon, the RDM killer.

Virtual Volumes (VVols) is a spec from VMware that allows storage vendors implement virtual disks as they see fit. On FlashArray, we’ve implemented VVols virtual disks as just regular volumes on the array.

Think about what that means for a second.

It means that you now get virtual disk granularity of VMs for not only data services on the array, like snapshots, and clones – but you also get virtual disk granularity for replication. You’re no longer forced to snapshot/clone/replicate entire datastores, or dealing with pesky, slow SCSI bus rescans and even more painful datastore resignature operations.

For a technical introduction to VVols, go ahead and watch this playlist with videos from my coworker and VMware extraordinaire, Cody Hosterman. Keep reading after for a discussion on running SQL Server workloads on VVols.

Done with the videos? Cool. If not, make sure you watch them later – they have the best explanation of how VVols work that I’ve seen out there. And even if you don’t use FlashArray, the explanation and concepts apply (mostly) the same.

SQL Server on VVols: The Good, The Bad, and The Ugly

Did I scare you with the title? Sorry, that was meant to throw you off.


With VVols, you maintain the performance of an RDM with the agility of a VMDK. No more worrying about thin vs eager zero thick or any of that nonsense!.

Moving between physical volumes, RDMs and VVols is extremely straightforward on FlashArray. A move to VVols from VMFS basically entails a simple storage vMotion, which on FlashArray is an array deferred, metadata-only operation. Since FlashArray is a data reducing array, a storage vMotion from VMFS to VVols is just a VAAI XCOPY operation. Your new “copy” will only consume a little bit of space for the metadata, and nothing else. An RDM to VVols migration is also pretty straightforward, and there’s no VMFS datastore “container” to worry about. It’s exactly the same layout on the array, whether you go physical, RDM, or VVols. And you’re not “stuck” with VVols once you make the decision to go down that route for a given virtual disk. If you need to go back to VMFS, just storage vMotion it back, simple as that.

When Can I NOT Use VVols Then?

There is only one scenario where you can’t quite use VVols for SQL Server, and that’s when you have shared storage between SQL Server instances, which is the case for Failover Cluster Instances (FCI), which you probably just call “clusters” and VMware keeps calling “MSCS” (for Microsoft Cluster Server), which is nomenclature from 2003, feels like 1989, and no one should be using anymore.

However, there’s good news on that front! VMware will support SCSI-3 reservations on VVols starting with version 6.7, which is in beta right now from VMware. Go ahead and read Eric Seibert’s blog post on VVols futures for more information on that. At that point, I don’t want any of you even mentioning VMFS to me, unless you’re running a version of vSphere prior to 6.5 (which is our requirement for VVols). But quite honestly, VVols is a pretty compelling reason to upgrade to 6.5, or 6.7 when it comes out!

So What About Cloning?

My coworker and former (should I say recovering?) Oracle DBA Somu Rajarathinam wrote a great post on Oracle database cloning with VVols and FlashArray. Rather than me just translating that post to SQL Server lingo, I’ll just link to it here and let you enjoy. The concepts on SQL Server are extremely similar so just picture in your head what that would look like.

That’s it for this post, but let’s be sure to continue the conversation on Twitter by mentioning me (@DBArgenis), Cody (@codyhosterman) or just using the hashtag #PureStorage.

Thanks for reading,


VVol Data Mobility: Data from Virtual to Physical

One of the most strategic benefits of Virtual Volumes is how it opens up your data mobility. Because there is no more VMDK encapsulation, VVols are just block volumes with whatever file system your guest OS in the VM puts on it. So a VVol is really just a volume hosting NTFS, or XFS or whatever. So if a target can read that file system, it can use that VVol. It does not have to be a VMware VM.

Let me start out with: YES our VVols deployment will be GA VERY soon. I am sorry (but not really) for continuing to tease VVols here.

This is one of the reasons we do not treat VVols on the FlashArray any differently than any other volume–because they aren’t different! So there is no reason you can’t move the data around. So why block it??

Some possibilities this function opens us:

  1. Take a RDM and make it a VVol
  2. Take a VVol and present it to an older VMware environment as a RDM
  3. Take a VVol and present it, or a copy of it, to a physical server.
  4. On the FlashArray we are also introducing something called CloudSnap, which will let you take snapshots of volumes (aka VVols) and send them to NFS, or S3 to be brought up as a EBS volume for an EC2 instance.

Continue reading VVol Data Mobility: Data from Virtual to Physical

What is a Config VVol Anyways?

I have blogged a decent amount recently about VVols and in many of those posts I mention config VVols. When using vSphere Virtual Volumes, VMs have one, some, or all of the following VVols types:

  • Data VVols–every virtual disk you add creates a data VVol on your array
  • Swap VVol–when you power on a VVol-based VM, a swap VVol is created. When you power it off, this is deleted.
  • Memory VVol–When you create a snapshot and store the memory state or when you suspend a VM, this is created.
  • Config VVol–represents a folder on a VVol datastore.

This statement about config VVols deserves a bit more attention I think. What does that really mean? Understanding config VVols is important¬† when it comes to recovery etc. So let’s dig into this.

Continue reading What is a Config VVol Anyways?

VVol Lightboard Videos

Quick post. I did some light board videos together on vSphere Virtual Volumes. Lightboard videos are pretty fun to do, the unfortunate part is that I have horrible hand writing. So I immediately apologize for that.

A common question I get with these videos is how do you write backwards. I don’t. I am nowhere near that skilled, as you can see I can barely write forwards. I write normally which appears backwards and the video team mirrors the video.

This is a three part series, the entire playlist can be found here:

Continue reading VVol Lightboard Videos

Moving from an RDM to a VVol

Migrating VMDKs or virtual mode RDMs to VVols is easy: Storage vMotion. No downtime, no pre-creating of volumes. Simple and fast. But physical mode RDMs are a bit different.

As we all begrudgingly admit there are still more than a few Raw Device Mappings out there in VMware environments. Two primary use cases:

  • Microsoft Clustering. Virtual disks can only be used for Failover Clustering if all of the VMs are on the same ESXi hosts which feels a bit like defeating the purpose. So most opt for RDMs so they can split the VMs up.
  • Physical to virtual.¬†Sharing copies of data between physical and virtual or some other hypervisor is the most common reason I see these days. Mostly around database dev/test scenarios. The concept of a VMDK can keep your data from being easily shared, so RDMs provide a workaround.

Continue reading Moving from an RDM to a VVol

Do thin VVols perform better than thin VMDKs?

Yes. Any questions?

Ahem, I suppose I will prove it out. The real answer is, well maybe. Depends on the array.

So debates have raged on for quite some time around performance of virtual disk types and while the difference has diminished drastically over the years, eagerzeroedthick has always out-performed thin. And therefore many users opted to not use thin virtual disks because of it.

So first off, why the difference?

Continue reading Do thin VVols perform better than thin VMDKs?

Comparing VVols to VMDKs and RDMs

I have been talking a lot about Virtual Volumes (VVols) lately with customers and when I describe what they are a frequent response is “oh so basically RDMs then?”. ..

…ugh sorry I just threw up in my mouth a bit…

The answer to that is an unequivocal “no” of course, but the question deserves a thorough response.

So first let’s look at how they are the same, then let’s look at their differences. And not just how they compare to RDMs, but also VMDKs as you traditionally know them.

Continue reading Comparing VVols to VMDKs and RDMs

My Upcoming VMUG Webinars

Hey all–in case you didn’t get a chance to go to VMworld and would like to see my sessions live (well online live) and ask questions I am repeating them somewhat via the VMUG webinars.

First one is on October 9th:

This is a deep dive on the love story that is ESXi and UNMAP. How it used to work, how it works now, how it has changed, and why. Plus how VVols changes it all. This is a session I submitted to VMworld, but it didn’t make it. Register here:


Second one is a repeat from VMworld, my session on best practicesfor vSphere for All Flash Arrays. Continue reading My Upcoming VMUG Webinars

Virtual Volumes: VVol Bindings Explained

Virtual Volumes change quite a lot of things. One of these is how your storage volumes are actually connected. This change is necessitated for two reasons:

  • Scale. Traditional ESXi SCSI limits how many SCSI devices can be seen at once. 256 in 6.0 and earlier and 512 in 6.5. This is still not enough when every virtual disk its own volume.
  • Performance. Virtual Volumes are provisioned and de-provisioned and moved and accessed constantly. If every time one of these operations occurred a SC SI rescan was required, we would see rescan storms unlike this world has ever witnessed.

So VMware changed how this is done. Continue reading Virtual Volumes: VVol Bindings Explained

VMworld 2017 Session Wrap-Up

I’m back from VMworld 2017 US and then VMworld Europe and a nice vacation. Time to get back to work and of course some (well a lot hopefully) blogging.

Had a great time catching up with friends at VMworld and talking about the new stuff that both Pure Storage and VMware have coming. I had quite a few sessions this year–VMware was kind enough to post online–many of them publically (no login required).

Continue reading VMworld 2017 Session Wrap-Up

"Remember kids, the only difference between Science and screwing around is writing it down"

%d bloggers like this: