Video: How Docker brought containers mainstream
Making a mixed estate of different pieces of IT infrastructure work together is a constant headache for many businesses, but the trend towards hyper-convergence means it can be possible to bring together old and new storage into one efficient hardware stack.
Virtuozzo is one company doing just that through a variety of software tools ranging from virtualised object storage to cloud-optimised Linux distributions to container migration technologies. Virtuozzo also works on numerous open source projects including OpenVZ, CRIU, KVM, Docker, OpenStack, and the Linux kernel.
Virtuozzo's CEO, George Karidis, recently told ZDNet how it's done.
As I understand it ,Virtuozzo is part of a group of company owned and run by Serguei Beloussov, the man behind Parallels?
Virtuozzo was actually the first company in that family and was created back in early 2000. Serguei had a view that, at that time, virtualisation in hosting wasn't working particularly well and that shared hosting was not very functional. He went off and built a container, at system level, virtualisation strategy and that became Virtuozzo.
We fundamentally created the virtual private server industry for the hosting community that was developing at that time.
The business has evolved since then and we focus on the service provider community. But that's not just hosting anymore but also includes other software components such as SaaS environments and even some systems integrators or managed services providers.
Our model, or software stack, has evolved to include virtualisation of the compute, virtualisation in software defined storage along with the networking capabilities to wrap it all together and create a cluster or stack around, basically, any x86 system-type infrastructures.
We can then manage storage across multiple nodes and provide the networking capabilities to bring it all together. Our model has evolved from pure compute at the container level to layering KVMs [Kernel-based Virtual Machines] as a virtual machine so we can actually, on the same physical node, let you run containers as well as virtual machines -- then build in the storage capability as a truly software defined stack that mimics, or from a definitional standpoint, is the same hyper-converged infrastructure structure.
Our difference is that we are pure software and so we can be hardware agnostic. And the systems can run on leftover hardware, or hardware that's been re-purposed, so the customer doesn't have to make a vendor choice.
And the same is true of the software on the storage side where it doesn't matter whether it is a hard drive or an SSD or any other form factor. We allow all of that to be managed through a common interface and a common software stack.
So you can run on virtually any hardware platform?
Correct. We are certified on any x86 platform as well as ARM. We do work there but it's just not a big market for us. Our OS is based on Linux and that includes Android and other variants.
Does that mean that hyper-convergence is a big part of your business?
Yes. It started as computer but people want to squeeze more out of the storage they already have and stop having to buy things like high-end SANs and pay the premium on that. A hard-drive is just a hard-drive, and depending on workloads and configurations we can be very competitive.
Our evolution is towards the hyper-converged and if you think of hyper-converged, it's fundamentally a storage solution. If you look at the mechanics, they are not really focused on the computer side at all, they are really a storage play and compute is secondary.
What's the difference between yourselves and the competition?
One of the biggest differences is that we are truly just a software stack. We don't care about the hardware. I think that gives us a unique value proposition in the market because the other companies tie themselves specifically to an appliance. They may come pre-configured and so perhaps are easier to use but then you are tied to that framework and that form factor.
Now, if you think of any datacentre environment, it's not homogenous. You may have Dell and HP or white label servers and you may have generations of that. We don't care.
Is your point of difference that you can get all this varied stuff working together?
Correct. Also we can bridge a virtual machine. If you look at a lot of deployments, especially in mainstream businesses, it wasn't containers five or six years ago it was VM. We can actually allow that to work and have a transition strategy where you can get things like Docker and the DevOps movement -- which is really about containers -- and then ship in edge-type solutions. We can deliver that on a common platform.
Do you think the move to containers is well established now?
The answer to that is: it depends. In companies without a legacy that are starting with containers and with a container strategy, it might take off but that's not the vast majority of enterprises.
I went to see the CEO of a security company recently and he's all containers and has gone with NGINX to enable that. That's one end of the market. But if you go to a mainstream enterprise that has two dozen workloads to run, then they're not all containers and they're not DevOps. At least not yet.
They may be talking about it. They may be pushing some cloud native apps into that model but that's not the mainstream. It's much like cloud was a few years ago.
How would you define hyper-convergence?
Our definition is: a software defined stack for data centre operations or just IT; it's not just storage. Most people assume it is storage but for us it's all three: compute, storage and network. That's what the convergence play is. If you don't have all of that it's nothing but storage.
You recently announced a dual-replication feature. How does that work?
A big area for us right now is our storage investment. Almost 60 percent of my development group is focused just on storage and that will evolve and the orchestration and management interface will evolve with it.
The biggest single announcement was the dual-replication. It allows us to do some interesting things around migration.
On the migration side, this is a way for us to help companies transition. It's not just about data centre transition but is also allows a shift to a DevOps model, or to a cloud infrastructure. It does this by creating a geo-replication system or solution that means that the customer does not have to shut down production but can simply replicate the data and point it to a replication recovery position. That becomes an interesting migration play for larger workloads.
Who are your largest customers?
We have some of the largest hosting and service providers like GoDaddy and 1&1. But we don't sell a lot of our product direct. The vast majority of our customers are service providers and that's a much more efficient approach for us.
The service provider community used to be the hosting industry, which then became the cloud services companies but is now evolving into a more mainstream VAR or system integrator market. Typically, for us, that is where the VARs have been selling hardware to their customers and the customers ask them about how they should address cloud. So then, typically, that brings up issues like, should they go private cloud or hyper-convergence? And that's where we are starting to see traction.
Now we are bidding a lot of projects around storage and virtualisation.
Do you think a lot of companies are going down the DevOps route?
I think yes, but the question is, how long will it take? Is it going to take off in the next 24 months? No, I think. Not in the mainstream, mass adoption way. I think it will be three, four or five years before we see it as a normal conversation among a traditional IT team.
Previous and related coverage
A10 Networks' cloud chief on apps, DevOps, security, 5G, IoT and more [Tech Pro Research]
Kamal Anand, VP of A10 Networks' cloud business unit, offers insights into cloud-native solutions and their role in digital transformation.
There used to be a bunch of vendors offering unlimited storage in the cloud. Not so much anymore. CrashPlan exited this space last week. What does this mean for the future of cloud storage? We take an in-depth look.
Advancements in storage have contributed to the progression of AI, even though it's still early days.