Today's leading minds talk Data Storage with Host Enrico Signoretti.
Jason Collier is a Co-founder of Scale Computing, Inc. and also served as its Chief Technology Officer until September 9, 2015. Collier is the Chief Evangelist for Scale Computing.
Enrico Signoretti: Welcome everybody! This is a Voices in Data Storage, brought to you by GigaOm. I am your host Enrico Signoretti and my guest for this episode is Jason Collier. He's co-founder and as he wrote on his LinkedIn profile, ‘Swiss Army Knife’ at Scale Computing. I hope that he will have the chance to explain what that means. Hi Jason, how are you today?
Jason Collier: I'm doing very good, Enrico, how are you today?
I'm fine. Thank you very much for being with me today. So [your] Swiss army knife [title] -- what does it mean?
I've had that on my profile for quite some time because I do a very broad group of things within the organization throughout my career, everything from things focusing on the business side to the technical side. I actually used to be the CTO of scale computing and I determined that that job was too difficult, there was too much work in it, it wasn't difficult, it was just too much actual work. So I gave that over to my chief architect, Phil White, and he's our CTO. I just randomly change my title to whatever it fits. I wanted Collier as a title just because, hey, everybody calls me that within the organization, I do a little bit of everything.
Very nice. So Scale Computing is a successful start-up in the hyper convergence space. You started a long time ago with a scale up storage at the beginning, which then evolved in what is now a solution that brings together storage and hyper-convergence. Like the vast majority of the competitors in this space, this solution is not based on VMWare, nor Hyper-V, but Scale Computing controls all the software stack and end-to-end. This solution is focused on SME and with the rise of edge computing, probably also the range of use cases is responding. Did I miss something here?
No. So our foundation, one of the things...it's funny because we kind of still call ourselves a start-up but we have been around for a decade now. So, I guess we're actually an established company, funny how that works right? So, like I said, we've been in the business for 10 years, we started in the scale out storage, but what it was, our whole plan the entire time, hence why it's a storage company named Scale Computing. Our whole goal was to get to what this word was, we were actually involved in the inventing of it - hyperconvergence. But the fundamental piece, the foundational component was for us to actually build a good scale out storage platform for the servers and virtualization stack to run on top of. In our original business plan, we had that happening, where we were going to integrate the hypervisor stack about six months after.
Well, it ended up being about four or five years after I think we founded the company that we finally got to that point of hyper-convergence, which I think was good because that was a pretty dramatic step from where things were currently running. You think of kind of the 3 2 1 model of the way computers have been done where you've got servers that are running, VMWare, Hyper-V that connect down into some type of shared storage, a SANs, so you can get those high availability features.
Our fundamental shift was, well, why not distribute the storage across everything, get that storage component really nailed down and then run that virtualization stack directly on top of that as well. Then you can use these commodity resources to kind of pull that off. And our underlying hypervisor that we use is actually KVM, but the thing is, you don't have to be a KVM expert to use it. One of our core design principles has always been simplicity. And in doing that, basically by the elimination of the SAN and the fact that it's running on top of these commodity X86 appliances that we sell, it then becomes an exceptionally simple concept to basically go in and deploy it, and then we also stack a user interface on top of it that is simple enough, that we actually had one of our systems engineers go home one night, took his iPhone, recorded a video, put it up on YouTube of his daughter implementing a Windows 2012 server.
The funny thing was she did that, that took her about 45 seconds to do on top of the scale platform, but she was four years old, and he paid her in chocolate chips. So it's one of those things where she didn't have to go through weeks and weeks of VMWare training to learn how to use the system. We designed it for simplicity so that you don't have to think about the infrastructure, you let the infrastructure itself worry about those infrastructure components.
On one end, you have simplicity, so ease of use is one of the most interesting benefits. I also think that support can be another benefit in this case because you control everything, so if something happens, you have control of the advisor.
We are the one back to pat and one throat to choke in a support scenario. If something goes wrong with the infrastructure, you call up our support. Our support is amazing, we've got basically 24 by 7 by 365 tech support. You can call us up, plus we also have implementations within the stack itself where if something goes wrong, you can enable a support tunnel that gets to our support team. So you can literally have people working on a problem in a matter of seconds versus a matter of hours to get a tech on site, plus the system itself, a lot of where our patented technology is, is around not only that storage piece but also the orchestration stack.
We have done a lot with the way that we do kind of collectors that feed into a checked value system that feed into this condition engine, much like when the check engine light pops on, on the car, what we've got the capability of doing, is going in and not only reporting on these errors, but also taking very complex situations and fixing them to create this self-healing style of environment. And I think we've got 27 patents or something wrapped up around just the way that we do that whole collection and the condition checking, and then feed that into state machines that actually take care of managing the complexity of a scale out distributed system.
Do the oldest benefits also reflect in the licensing model, meaning I want to buy a box or a number from your machine and you'd give me what's necessary?
Yes. And everything is included, so it's basically all the hardware, all the software, and your premium tech support is included and bundled in right from the start. You've got a single support model, and we also do not do a la carte licensing. I won't mention the name, but I had a number of storage products in the past when I was running basically VP of operations for a couple of other organizations in the IT field. But every little thing that you needed on the storage... ‘oh, you want to SMB? Well guess what, that's an extra licensing fee. Oh, you want this? Well, that's an extra licensing fee. You want to connect that to this? Oh, that's an extra licensing fee.’ And we are vehemently opposed to that whole ‘license every feature’ [approach], so not only everything that we've got, but everything we're coming out with is included and bundled in.
One of the interesting things that we're going to have integrated into the solution here over the next few months is the ability to even have a level of file level restore that's actually built in. So if you have relatively simplistic backup needs, you can have a backup system where you can go down to do file level restore, and that costs our customers nothing extra.
At the beginning, we also talking about the fact that SMB is your primary market, so there's more medium enterprises in general, but actually, with edge computing you are discovering new scenarios where the solution is good for it, right?
Yeah, it's a phenomenal fit, and the funny thing is, this whole concept of edge compute has come out and you've got a lot of companies talking about the future of edge computing, and the reality is, this is what we've been developing the entire time. So while everybody else is out there talking about the future of edge computing, we're the ones building it, and have been for quite some time. And some of those really good use cases for edge… an example of this: we have a large retail chain that's based out of Belgium. They have a total of 8,000 stores. What they're running is, we've got this effectively micro data center that we are implementing in all 8,000 of those stores. So, when you look at the individual requirement, it's very much kind of like an SMB-SME requirement for those individual stores and that whole simplicity. They don't have IT staff at every retail location.
What we've got is this ‘data center in a box’ where you effectively roll it in, and we've got some even newer things, and I'll get kind of to this in a second when talking about efficiency of the stack. But when you need to go in and deploy something, you can have literally a grocery store manager basically put it in, plug it in, turn it on, and it effectively boots up, it's part of the environment. So you don't have to have anybody with any IT training go in and do this field implementation of these systems.
And that is an exceptionally good kind of scenario where this edge compute, when you think about it, it's no different than that SMB and SME style of deployment, it's just the fact that there's a lot of them. And we've also built the kind of integrated construct that helps you manage that many locations. So from that edge perspective, we've got the capability of, I think we can manage up to like 9,000 individual locations with the current setup that we've got -- all from a single pane of glass.
This is interesting. So from the central IT organization, you can have a view of all these locations, the different machines that are running an application, and you can access the system remotely if you need it?
Yes. And we also have our built-in snapshotting replication, so say their point of sale system or their inventory control system can be snapshotted and replicated up to corporate.
So, you've got this idea of getting disaster recovery for a single location up to the data center?
Yep. And you could also have those replicated to other clusters in the region. So, if there was another grocery store that's like 20 miles away, you could have certain contexts of, say inventory that's salvageable and they could like move it over to another store where they can have that inventory system also replicated to something that's actually closer. So, there's a lot of different use cases for it, and a lot of the technologies that we've built in have become pretty attractive for those edge style deployments.
I understand. And your model is always to sell the appliance right? You don't sell the software and the customer buys the hardware?
Well, that's kind of a situation where that's changing a bit. I would say one of the areas in which that's changing is with some of the partnerships that we have developed. An example of this is, we've got a partnership with Lenovo, specifically around doing this deployment, in fact, that grocery store chain that I'm talking about, that was actually a joint deal that we did with Lenovo. Now, the reality is, there is nothing about our hardware that is special. It is in fact as commodity as it can possibly get to the point where that storage stack, that we explicitly built it, so that you didn't have to have an expensive RAID card to get RAID level redundancies within it.
So, we basically built that software defined storage layer because we wanted to make those X86 boxes as commodity as we possibly could, as low cost as we possibly could. And all of our intelligence is built into the software. Is Scale Computing a software company? Yes, we are. That's what we do. And we deliver it in an appliance model too for that simplicity component. The real reason why we always delivered as an appliance was to basically give almost that iPhone level of experience, where you pull it out of the box, it's not like you pull an iPhone out of the box and then you get an SD card and you snap it in and got to install the OS and things like that. Our deployment model was always appliance focused simply to give the customer that great experience.
Now that said, when you start getting into things, in another area where we've had really good success is things like managed service providers. And we're getting to the point where, say like a managed service provider has a bunch of end customers, they want to sell a single node scale systems out to them, and then they can be the disaster recovery point for it. That's when we get into looking at, yes, we can basically modify our stack to run on other commoditized hardware platforms.
We are a software company, and we're getting to a point within specific markets where we are selling it as software, but this is one of those things where if an SMB comes up to us and says, ‘hey, I've got a couple of supermicro servers here, I've got a couple of HP servers, can I run those, run it on that?’ The answer is no. And one of the reasons is, what we find is -- and I'm sure you've been through this so many times too in your career, Enrico -- some bug could be caused by when the drive firmware doesn't match up or there's some bug with this firmware and the RAID card that then has a driver issue in the operating system, and in going in and controlling the hardware as tightly as possible causes so many fewer support cases to actually happen.
Going back to that support, that's one of the reasons why...Our NPS or net promoter score has consistently been in the 80s and 90s. I think the latest quarter that we measured our net promoter score was 92, which is exceptionally high, especially in the IT field. And that's one of the advantages. I think in general you'll see most hyperconverged companies in general do have very high net promoter scores because, like what we said before, it's the ‘one back to pat and the one throat to choke.’
Do you sell these appliances directly or do you have a channel?
We are 100% channel. We sell through a two-tier distribution model and depending on the country that you're in, we have distributors pretty much across Europe, Africa, and honestly, the Lenovo relationship has really expanded that as well. In going through and selling these edge deployments, we've been doing very well on the African continent. And Lenovo is effectively acting as the hardware support partner on that. We've also done very good with Lenovo in Europe in general. So that's been a very good partnership. We just announced it, but the reality is we've been doing a lot of fieldwork with them, and also utilizing their channel to help scale, expand on a global basis. But, yeah, we're 100% channel. We do not sell anything direct.
So, maybe this is a good segue because I wanted to ask you about an update of the company from the growth point of view.
Yeah, so the company has been growing very well. We recently received a $35 million round of funding with a variety of investors involved in that, including multiple other strategic partners. And that for us segues into the next growth stage for scale, so we're really going through and expanding. The product is exceptionally solid. Now we're going to continue to invest in R&D, but we're really working on expanding out the sales of marketing channels that we have got with this additional tranche of funding. The good news is this tranche of funding is going to put us into something that's pretty unique within the industry. It's pretty unique within the tech industry, but very unique within any type of storage or hyperconverged industry, is this round will take us into a state where we can say we are a profitable company.
So this will take us to profitability, and that's one of the things that Jeff, our CEO, he and I are kind of passionate as well as Scott Loughmiller, who is our VP of engineering, but the 3 of us, as the founders, one of the things that we want to do is to create a longstanding company within the industry, and I think this is going to position us very well to actually become a company that can stand on its own legs with basically the sales that have been ramping. With that, our customer bases have expanded. I think we've about 3,500 customers at this point and we've been experiencing fantastic growth, just quarter over quarter. And what this additional round of funding is going to do is going to help us grow even more.
And then, we've already talked about that partnership with Lenovo. That helps us, like I said, expand the channel and get into territories that traditionally would be difficult for us, given the size of the company. Now even given the size of the company, we're a relatively small company compared to the number of customers we've got. I say 3,500 customers, but basically, we're not even 150 employees yet. But that said, we're pretty much on, I don't want to say a hiring binge, but that's kind of what it is, and it's really for scaling out the support operations of basically expansion. The growth has been phenomenal, but we also have been trying to do growth in an intelligent way, not just pumping a bunch of money and hiring a bunch of people everywhere and growing it that way. We are trying to be intelligent about the way that we are expanding the base.
That's great. How can we continue the discussion? I know that you are pretty active on social media for example, so can you share your Twitter handle or the company’s?
Yeah. The company's Twitter handle is Scale Computing, @scalecomputing, and mine is, everybody always asks where this came from, and I think I've always told people...I'm just like, we'll have to have a long private conversation for me to actually reveal it, but my private Twitter handle that I tweet from is @bocanuts.
Ok, we'll ask you later. Last but not least, is it possible to try the product? Is there a demo of it?
Absolutely, you can go to our website, which is scalecomputing.com, and you can go on there, and you can directly from there, we've got a live chat feature on there and there's also a requested demo button right on there, so you can go and request a demo and we'll have our team get back with you.
Ok. Jason, I think we covered a little bit of everything today. Thank you again for your time here at Voices in Data Storage. Bye-bye.
Thanks so much, Enrico. Good talking to you.