Interview

Mike Koponen- Pivot 3

Mixed Application Workloads in Hyperconverged
Infrastructures with Pivot3

Mixed Application Workloads in Hyperconverged
Infrastructures with Pivot3

We have seen HCI growing and expanding, it’s definitely evolving, it’s not a settled market at all. Part
of the reason you might adopt HCI is because you want to simplify your IT stack. You might want to
buy modular infrastructure. You might want to take the expertise that you have and focus it on
something other than integrating disparate technologies.
But, when you build that HCI stack and get that piece into your data center, you still run into
issues of policy based control of the resources.

You still run into issues of capacity. You still run into issues of availability, noisy neighbors, and my
favorite problem, performance. Pivot3 has got some technology that they’re applying to this.

Can you just describe to us what Pivot3 has done in the HCI space with your Acuity Launch? What
have you brought to the table?

Customers have latched on to hyperconversion for searchers to take and be in where a simple, less
complex way to deploy servers and storage, and provides them a lot more agility, but with that
comes the challenge of how do you support multiple mixed application workloads on this shared
infrastructure that you’re both running virtualized applications on, and using shared storage. You’ve
got to ensure that it’s got the performance to support all those mixed application workloads including
mission critical ones. And you need to be able to take in and eliminate the contain ship of resources
that’s going to exist. So, when we launched our acuity hyperconverged software platform and
appliances, we take and make use of NVMe flash in those appliances, so that provides us ultra-low
latency, high-performance of front-end on those appliances, and then we match that with this
advanced policy-based engine that takes and has the ability to take and prioritize the more mission
critical applications over the less critical ones to ensure that they get priority access to resources and
thus, their performance SLAs are gonna be met. So, it’s the combination of this policy-based engine,
automate and simplify all of that performance management with this NVMe flash data pad that
provides the ultra low latency for those performance sensitive applications.

So, basically, in HCI, we put all our eggs in one basket, so to speak, when we put this modular
infrastructure up. And if the solution just can’t discriminate between workloads and it can’t optimize
the resources and align them properly, and effectively provide what we might call cu lass, to each
workload and really ensure that all the way down to the storage layer, you’re gonna get a whole big
box of contention. It’s just something that’s gonna be a problem.
And that’s why early on, you saw hyperconverged infrastructure being deployed more towards
targeted siloed use cases, like VDI as the poster child for hyperconverged early on, where a
company was going to take and stand up a virtual desktop infrastructure, and they said, hey, we’re
gonna deploy a separate infrastructure for that because we want to ensure that those virtual
desktops have the adequate performance that they need to be successful. So to guarantee that,
they brought a brand new infrastructure just to do that. Or a company said, hey, we’ve got this
database environment and this business analytics environment that sort of stand alone, it’s a
departmental type thing, we’ll do hyperconvergent infrastructure for that. So customers have had
success with that, and they got a taste for the value of hyperconverged and thus, they’ve been

asking more, hey, we want to support more workloads on this and consolidate more, but we need to
ensure that it’s going to deliver the performance, take care of this resource contention. And then
have all the data services we need and so that’s what we delivered with the acuity hyper converged
platform.

Another issue that comes up with hyperconvergence that we see today is that you’re still buying on
premise infrastructure. And you’re still managing a data center. People are starting to think instead
about how do I move some of this to a cloud? How do I build at, go all the way to a cloud? But, if I
still have a need for on premise infrastructure, a lot of people do how do I build a hybrid solution that
spans that? What are you doing in that space? I know you’ve got some announcements you’re
making.
We do indeed see what you described in terms of more companies want to take an Augment, their
on premise infrastructure with resources available in the public cloud. Particularly in the area of
backup, archiving, disaster recovery, and then we see some initial desire to take in actually push
workloads out into the cloud as well, where customers need some burst capacity from an application
standpoint. Or they want to spend up a quick dev ups out in the cloud. So what we at Pivot3
announced last month in March, was we launched the Pivot3 intelligent hybrid cloud engine. That
builds off the native intelligence that we have in our hyperconverged software platform in that area of
policy based management, dynamic resource orchestration that we spoke about earlier. And the
very first instance of that that we brought to market is what we refer to as Pivot3 cloud edition for
Amazon web services, AWS. And so, we’re leveraging our data replication engine, all of the policy
based management capabilities, to very easily allow customers to take and do backup to AWS. And
it’s a neat inflection point for us, in that as part of doing this, we now have the ability to run our Pivot3
hyperconverged software platform in AWS natively as an Amazon machine image and then take in
and do all the orchestration and management from a central Pivot3 management application. And
so, it’s very easy to set up. Very easy to take and manage for customers who want to take and
leverage AWS for backup.

What does it really means to have hyperconverged infrastructure in a cloud ? Because it’s already a
cloud, or because it’s already hyper converged, or whatever. But, I think the main point is I can build
a hybrid architecture where I can run mixed workloads on premise. I can burst to a cloud. I can back
up to the cloud. And I think as we were talking before, I can even recover then on the cloud’s side
from your backup point time snapshots and everything else you would have up there, right?

Yes, no exactly. To take off to a point that you made earlier, there is a lot of evidence that
hyperconverged infrastructure is an ideal platform to base a cloud on, because we do have a lot of
cloud service providers. Regional based cloud service providers, who are using Pivot3 as the basis
of their services offering. Where they’re providing backup as a service, a DR as a service, things of
that nature. And we’ve done that to date through alliance partner relationships we had with
companies like Zerto, Embeam and Viamware, so leveraging their tools. And so, what we’ve brought
to market now, with this Pivot3 cloud addition for AWS is a native solution that leverages the policy-
based management, the engine, the nanic resource orchestration.

Which comes first, HCI or the cloud? Or the cloud or the HCI?

Right, which builds on which? Which is great. Yeah, I mean if the service provider is concerned
about anything, it’s how do I host mixed workloads? Because by definition, this has mixed clients
and mixed workloads and without any way to really do that, they’re building up their whole big stack
of very fragile kinds of things. Right, now you can come to then and say, look, I’ve got basically
modular building blocks to basically build your cloud out of that comes up that policy-based stuff, that
comes with that orchestration and you can deliver that.

The other thing that you made me think of is that either cloud service provider or the IT organization
that’s providing those private cloud services. The other thing they have to be able to do is to
guarantee service levels to their user community as well. So the policy-based management, the
quality of service capabilities we have for our performance and data protection standpoint, allow that
cloud service provider or that IT organization to very easily take and do that without having to be
storage performance experts

Comment here