Primary Storage


The Big Picture:

The convergence of it primary or secondary...with services designed to protect that data is a huge benefit today. Toss in a mix of on-prem, off-prem, public and private cloud...with the ability to seamlessly protect and access that data across all, from a single console, is also a game changer. In this short video, learn about ClearSkyData and their comprehensive storage, data protection and disaster recovery solution.




Mike Matchett:                  Hi. I'm Mike Matchett with Small World, Big Data. I'm going to talk today about some of the most interesting trends in storage. Talking about the convergence of primary and secondary storage, the inclusion of storage and data protection, and cloud, multi-cloud storage, cloud storage and a couple different facets. I'm going to have Laz Vekiarides from ClearSky and we're going to talk about all these things.

                                                      First, I just want to mention a little bit about some of those trends I mentioned. When you're looking at storage, you probably got a bunch of initiatives going on. You've got things about end of life for your primary storage. You've got concerns about really getting good DR to cover your entire environment. You've probably got expanding license cost for your cloud and your backup software. How are you going to get a handle on that? Are those five different initiatives, or are they one? I think we're going to hear today that you can take care of a lot of those initiatives with one solution that bundles everything together in a very nice way. Welcome to the show, Laz.

Laz Vekiarides:                   Thanks for having me.

Mike Matchett:                  I should mention Laz is CEO and co-founder of ClearSky, and knows everything there is to know about storage. I actually used to one time work for a very brief amount of time with him at a former company not to be named in this show. First, Laz, what is ClearSky up to? What's kind of the shape of ClearSky Storage Solution that makes it different from everybody else?

Laz Vekiarides:                   Well, so we're a service, and we bundle, as you said, the convergence of primary and secondary. We bundle all of the capabilities that you'd expect from a primary storage system, along with all of the data protection and business continuity features that you'd expect to get from secondary storage, and all of the software that's associated with that.

                                                      The way we do it, which is unique in the world, is that we leverage the public cloud. We provide a single durable copy that's in the public cloud, for pretty much every byte of data that the customer has. That single durable copy being in the cloud, it has 11 nines of durability. It takes away a lot of the management that customers always have to do with physical storage, disk drives, et cetera, going bad. You don't see it. It's all in the cloud, and as most public clouds are built, it's several different buildings in a very secure facility offsite. So that's one tenet.

                                                      The other thing that is unique about us is that we run this service out of our own private network. And so connecting to the cloud and getting data in and out of the cloud, is a solved problem. We have very, very low latency, very, very high throughput. Essentially, our customers get the experience of an all flash array in their data center facilities, regardless of where they are. They also have the flexibility of using compute anywhere our network exists.

                                                      We're up and down the East Coast, and we're branching out into the Midwest right now, but over the next year or so, we're going to be coast-to-coast. You can have a colo facility, a compute facility, anywhere, plug into our network and access your data from there, which gives you a lot more options and a lot more flexibility with respect to business continuity, DR, small office, remote office management, all sorts of things that could be leveraged there.

Mike Matchett:                  When we talked over the years, you guys had the performance, because you did this close networking, what we call metro networking, metro area stuff. And now you have the durability because you're using the cloud, so you brought that into the puzzle for people. I really like the one durable copy, and it's probably worth emphasizing that, that's how you buy it as well.

Laz Vekiarides:                   Sure.

Mike Matchett:                  People don't pay for multiple copies all over your network, they pay for the one copy of storage, so it's really easy to understand what the cost structure is, and that no matter how many times it's going out there, it's one copy and it's a master copy. So that's pretty cool, but there's a difference between durability, and back-up, and DR. There's a big gulf between that. When you start to do DR, you have to orchestrate and automate stuff. You have to bring things up one place, bring them down another, be able to do, sometimes, fairly spurless replication. I don't know if I'm using the right word there. What did you guys mean when you said you can do DR, and how are you differing there?

Laz Vekiarides:                   Well, so you don't need to have an extra silo of storage in a secondary location. That's for starters right? So that's the most obvious thing. That's a huge amount of expense.

Mike Matchett:                  Back to the one durable copy, right? You're just leveraging that in a we are?

Laz Vekiarides:                   [inaudible 00:04:26] do that. Secondly, there are no software licenses for the orchestration software. It comes built in with our service, so if you're running VMware, we will give you our [inaudible 00:04:38] plug-in and it plugs into our network, it plugs into our edge infrastructure, and you can do all the orchestration from our console. And it understands all your locations of all your data centers, and you get the ability to do this orchestration out of our software. So you don't have to worry about buying another software package, and the amount of actual orchestration that used to be done manually, is actually relatively small. And over time, it's going to be de minimis because we have it as an ultimate goal that we're going to be one touch regardless of your application, regardless of your host operating system.

Mike Matchett:                  But DM's not just storage, it's about compute too. I have to bring up my competence there. And I know you guys have got some announcement you just made recently on that. How's that fit in?

Laz Vekiarides:                   So, one of the things that has been really tantalizing since we started the company was the notion that maybe somewhere, somehow, we could leverage on-demand computing. Like there'd be some compute somewhere that a customer would be able to rent and just orchestrate DR, without having to buy it and stand up physical presence in a data center somewhere. So, in the cloud area, you would think that public clouds would do that. It turns out that there are actual VMware partners that are service providers that will provide us on-demand compute services in a natural VMware server, or a VMware cluster.

                                                      So what we recently announced is a partnership with Faction to deliver an end to end DR solution, where Faction provides the on-demand compute. They literally will provide you with VMware instance, they call it a Faction Block. And it is big enough to run all of your lights out, pilot light services for DR site. And the cost is very, very contained, and this is actually right sized for something that doesn't get used very frequently. They have a budget built into that cost so that you can run a couple of tests of your DR infrastructure every year. And then when DR is actually a disaster recovery process actually declared, they will super size that instance. They will actually make it large enough to hold all of your compute, and you can just bring over all the data storage that you have running in your primary data site with us. You can literally just move them over into the Faction cloud facility. And within a few seconds, you'll be up and running.

                                                      So this is true on-demand DR. The cost is di minimus for the customer, especially when you compare it to all the physical infrastructure and the colo cost and power cost that they would have ordinarily. And it's very, very easy to use, and as an added value, our partner Faction will actually nag the customer to do a DR test twice a year to make sure that the stuff works, and they've actually thought through all of the of problems.

Mike Matchett:                  And if you don't agree, they come by and cut the power to your building anyway, right, or something in the middle of the night. So when we look at some other solutions out there, public cloud providers, obviously some places run VM's in the cloud nowadays. They still just can't get their storage from one place to the other overnight. You guys have literally zero cut over, right, zero RPO I think you were saying, because your storage is already there. That's the durable copy, so cutting over to your fail over site is like that.

Laz Vekiarides:                   Yeah, it's caching behavior, not tiering behavior. We have super fast networks, we have all sorts of optimizations. Our core intellectual property, you'll recall, is centered around the notion of building multi-tiered caches, and then deciding where to put things based on latency profiles, rather than just performance profiles. And so, all of our intellectual property lives in those layers and we're able to show very, very low RPO's, and we're also RPO zero inside your metro. When you are secretesly populating our network with your data when you're doing rights. So, not every single byte is protected by being off site. So it's a really interesting solution, and it brings new levels of protection. And I think with this additional partnership, we're completing the entire end to end solution.

Mike Matchett:                  I'm fascinated. There's lots of people out there looking at their end of life storage arrays and going, "What do I do next?" They're looking at initiatives for, like I said, back-up. They're looking at, "How do I control my license and costs on those secondary storage device? How do I control my cloud costs on storage that I moved up there? How do I start to take advantage of all this stuff?" And you guys have wrapped to up into one just consumable storage service, which is great. The final question I wanted ask you a little bit about was, we keep saying "cloud" and we keep saying "one durable copy," but it's really a multi-cloud thing I understand, right? You guys are just talking about one cloud, you're actually saying, "Hey, we also can solve your multi-cloud problem."

Laz Vekiarides:                   That's right. We run a very, very big fat fast network, and it's interconnected into all of the public clouds. So, you don't have to ... If you want to choose one, we will let you choose one, and that durable copy that you're paying for, it will be in that one. Alternatively, we'll pick one for you, but either way, that's not a decision that's going to kill you in the future. You can access your data across these different clouds because we have the network, and because we have these cloud proximal sites.

                                                      And our architecture with edge cache is, the software only, will allow you bring up in any location, a software edge. Essentially even in the public cloud, we're able to bring up Amazon, for instance, a [inaudible 00:10:33] in a VPC and you can attach to your storage using Amazon instances. So, that same technology is available across all the clouds, and you don't have to be burdened by your decision to use one cloud or another, especially if you've suddenly realized that the tools you really like are in another public cloud.

Mike Matchett:                  I mean I put everything in Amazon, but I want to use [TensorFlow 00:10:58], or do something else, but it's there. The data can be there, just echoed across. Awesome. Well I think that's all the time we have today Laz. I learn something every time I talk to you about the storage industry. We got to talk more often because I need to learn more things.

Laz Vekiarides:                   I learned a few things too, so this is great.

Mike Matchett:                  Thank you very much. Finally, I assume the website's got everything you want on it. Could you tell us about that?

Laz Vekiarides:          And pretty complete, you can reach out to us. We're happy to talk to you about any type of storage management problem you have, and love to show you how you can use the cloud and opX to solve all these issues.

Mike Matchett:                  Great, well thank you for being on the show today, Laz.

Laz Vekiarides:                   Thank you for having me.

Mike Matchett:                  And thanks for watching. I'm sure we're going to hear more about ClearSky Data as they get closer and closer to being the one solution you need for storage no matter what it is, as I keep hearing, and we'll be back soon. Thanks.

Laz Vekiarides:                   Thanks.