Ask the Experts: SPBM and VVols Storage with SolidFire

In this whiteboard video Josh Atwell, SolidFire Cloud Architect, and Rawlinson Rivera, VMware Principal Architect, dive into the details of how SolidFire and VMware are changing the way applications are deployed in vSphere environments. To learn more, visit solidfire.com/vmware.



Video Transcription

Josh: Hi I'm Josh Well, Cloud Architect for SolidFire and I'm here with my friend Rawlinson Rivera from VMware. Rawlinson today we're going to talk a little bit about SolidFire and we're going to talk about VVols and some of the technologies that enable virtual volumes in the latest release of vSphere. Let's start with first and foremost, what is VMware vSphere virtual volumes?

Rawlinson:VMware vSphere virtual volumes is basically new management framework that is being introduced into the platform to take advantage and eliminate a lot of the pain points that exist today when it comes to managing and dealing with storage and being able to deliver a fine granular kind of approach to the way we want to deal with our applications in terms of their performance and their values and their characteristics.

Josh: Great, let's start with kind of talking about some of the components that enable VMware virtual volumes to exist. I think from the SolidFire perspective the first major addition that we have to be concerned with is our VASA provider. The VASA provider in vSphere 6, in specifically related to virtual volumes enables our system to present up to the VMware environment, the capabilities of every volume that's located on the system. But from a VMware perspective you guys have opened it up to allow VMware to then leverage the VASA provider to execute all calls. Whenever you deploy a new virtual machine you make the call to the VASA provider to create the respective volumes that match up with the disks on the virtual machine.

Now on the VMware side what is happening on the VMware side to enable consumers to deploy that virtual machine and leverage virtual volumes?

Rawlinson: One of the things that happens with VASA, the version that's utilized in virtual volumes is VASA 2.0. In number one it enables a two way communication between the array and us. In a previous version, which is a one way communication like you said, it exposes capabilities. Now we're not only able to see the capabilities but we can tell the array how we want to consume those capabilities. The way in which that's managed, obviously the VASA provider will talk between nothing will connect those two. But we need some sort of way, some sort of construct that'll allow us to sort of tell the system what we need. In the way in which we do that is by basically we have a policy framework in which we call a storage policy based management. This framework allows us to take the capabilities that are exposed from the array, present them to vSphere and within capability there is a subset of values that can be identified and chosen for a particular volume or application.

Doing so you're able to create these sort of policies and then assign them to the virtual machine. Once the virtual machines are deployed on each particular virtual volume depending on their characteristics and also their policies, we can then tell the array or a set of APIs how we want to consume it, what kind of profile they want and how they want to get to that point. But essentially it gets to one container. We don't have to dealing with the precious metal and the silos architectures that we develop, where we have gold, silver, bronze. They tell about not being able to ask specifically for what type of capability performance profile you want, guarantee IOPS. We're able to kind of provide that to the individual volumes and then SolidFire kind of takes care of that down below.

Josh: From a SolidFire perspective, what we end up doing is in that metadata that is presented up from the VASA provider, we provide things like minimum IOPS, Max IOPS and burst IOPS. Also, if we have things like encryption at rest on the cluster that the storage container is a part of we can present those up. They can then be leveraged by a storage policy based management. As a result, what we're able to do at SolidFire is ensure that when that virtual machine is deployed, if this is an application disk and let's say that this is a database disk. Those two have very distinct personalities and profiles and requirements in order to provide the functionality that, that particular disk needs. Then what we're able to do is using storage policy based management and defining each individual volume to that characteristic. We can map those directly to the appropriate performance in profile.

Then on the VMware side, if we deploy out with SPBM how does VMware and SPBM ensure that all of that stays the way it's supposed to?

Rawlinson: Well now in vSphere there's a constant of compliance. It's part of the source policy based management framework as identifying whether or not there's a compliance factor on the values that are being delivered through that policy. If and when there's something that goes wrong whether it's a performance degradation, whether it's a failure on the device, whether it is capacity. It's not being able to satisfy what's being guaranteed as a policy up here, we'll be able to track that and be able to provide individually not per VM but per object individually a noncompliant alert. Which will let you obviously know that somethings not being satisfied and at some point in time either the issue has to be addressed down here by increasing or something doing that will deliver the guarantee resources that we're asking for. Then the system itself will sort of do the self healing mechanism of bringing you back into the compliance factor where things go back to normal.

Josh: Right, which leads to a great deal of capabilities to joint VMware and SolidFire customers to allow them to implement this technology to ensure that every application, every disk is getting exactly what it needs, no more, no less. Then being able to maintain it and keep an eye on it over time. As you'll see in a later video, we'll also talk about how we can move that upstack to use additional capabilities with virtualized automation and other technologies. Rawlinson I appreciate it. Stay tuned for some more videos on the topic. Thanks for watching.

Types: Videos

Categories: VMware, Product,

request a demo
contact sales