>> Announcer: Live from New Orleans, it's theCube. Covering VeeamON 2017, brought to you by Veeam. >> We're back at VeeamON, Dave Vellante with Stu Miniman. Marc Crespi is here, he's the vice president of SEs at Exagrid Systems, big partner of Veeam's, big presence on the show floor here. Mark, thanks for coming on theCube. >> Thanks for having me. >> So what's doing with Exagrid, we were talking off camera, kind of know you guys a little bit, you guys are right around the corner from us in Massachusetts, but give us the update in the company and what's new? >> Yeah, be happy to. So, first I'd like to thank Veeam for putting on a terrific show, and it's great to be in the beautiful city of New Orleans with you guys. So, if you look at the Exagrid business, Exagrid is a leader in disbase backup with data deduplication business. And we've been a Veeam partner for a decade now, and right from the early days when we started talking and working with Veeam, we realized that our two architectures had a natural fit. So when we talked to joint Veeam customers, whether they're new customers or existing customers, they're experiencing an exponential benefit over just using Veeam with some other disk player as result. If you look at how our business has evolved over the last decade or so, we were originally in the tape replacement business, you know, the dinosaur tape libraries that were still roaming the earth back then, and what we find now is, a lot of customers have moved on from tape, tape is a minority of the backup storage media that we see in the market today. And most of our business in fact is replacing other disk based implementations, either with or without native data deduplication, about 80% of our business now. And it's all the names you hear in the disk based backup with data deduplication market that we're replacing. We've also grown from a company that initially focused on midsized enterprise to now an enterprise class built product and company. So if you look at our average sale, our average customer size, it has grown exponentially over the past several years. And our sales force has grown over 500% just in the last two to three years itself, so we're in a high growth mode, we're experiencing a lot of success and much of our business is, a significant portion of our business is working with either existing or new Veeam customers. >> And a lot of the growth is coming from replacing existing, what's generally referred to as purpose built backup appliances, is that correct? >> That is correct, and the reason that we're seeing that phenomenon is when we sat down and created our architecture, we looked at the legacy of tape and what was wrong with tape. Well tape wasn't very mechanical, it was unreliable, but it also suffered from a vicious cycle of grow, break, replace. So, all our customer data is growing 20, 30% a year, which means your data's doubling every 2.5 to three years. And whatever you're backing up to, you're going to outgrow it. And you're going to ultimately have to replace it in its entirety. And you've got those precious IT budget dollars that you'd like to spend on other initiatives, and you're rebuying your backup storage just to tread water before you even get around to spending on the expansion. So we said that problem needs to be eliminated entirely, and the only way you can eliminate that problem, is by having a highly scalable architecture that never requires forklift upgrade. So if you look at our technology and why we're able to replace incumbent vendors, we're typically finding a frustrated customer who's been through two or three forced refreshes, either 'cause they outgrew technology or the vendor forced them to to outgrow technology by end of lifing et cetera, which we don't do, we don't end of life any of our products, and therefore they lift their head to say, well before I just spend all these dollars again, plus expansion, why don't I go back into the market and see if anyones figured out a better way to do this, and that's where we come in. We come in and show them that you can start with the footprint you need and then you can expand infinitely and we're never going to force you to buy what you already own, so, it marries up much more closely with the lifespan customer customers want for backup storage than the lifespan vendors want for back up storage. >> Marc, can you unpack that a little bit for us, I think about VM where it was an example of how we avoided having to do certain upgrades. I think of operating systems, or servers that were end of life, stick it into VM, I could grow and expand, but when I think about Gear, there's all sorts of reasons why just the exponential growth of, you know, different media types, different sizes that we need to take, that how come you can do this, while others, you know, force those upgrades. >> That's a great question, so I'd compare and contrast a little bit with virtualization, what virtualization brought to the table was, it allowed you to take a set of computing resources and make sure it was fully utilized, right, so if you had a server, you were running one application on it, maybe it was only 30% utilized, you had spare storage, you had spare compute, so what virtualization allowed you to do was add applications that were segmented, and therefore they could run without conflict and you could get that hardware fully utilized. This is a little bit different in that, if you think about what backup really is, on a nightly or weekly basis, even with some of the modern backup techniques that have come out, customers are moving large amounts of data, and it has to be within a certain window of time, because they don't want backups running during productions hours, because that can impact network performance, server performance, et cetera. The other side of the equation is when they want something back they want it back fast. So in order to achieve that, we made two architectural differences, on a scalability side, we said that the legacy storage architectures that typically, utilize a fixed amount of compute, and then expand by simply adding storage, missed the point that when you add workload to a system, but you don't add power to that system, performance at the same time, everything that system does is going to take longer. So, if I have a certain amount of data, and I have a certain amount of compute, and then I double my data, but I don't double my compute, my memory and my networking, naturally everything that system's doing going to take twice as long. So we recognized that you needed a grid based architecture, or a cluster based architecture, that said, when my data doubles, I'll double the storage, but I'm also going to double the compute, the network, the memory, et cetera, at the same time. So if I have a very short backup window day one, with an Exagrid implementation, and my data doubles, I have that very same backup window, I have the very same recovery time. I have the very same replication time, all the things that a disk based backup appliance do, grow linearly with Exagrid. >> And you're saying other architectures had to wait for intel? >> That's a great point, yes, they rely very much on the compute. Now there's implementations where Flash is being added to try and speed up processes, et cetera, which sounds like a great idea, 'cause Flash is obviously a very useful technology in the storage industry, but when you look at the pricing of backup infrastructure, Flash breaks the model for that, for backup infrastructure. It makes the products more expensive and its unnecessary if you implement things correctly. >> Because FAT Disk is still cheaper than cheap Flash, is that right? >> Spinning disk is still about a sixth to an eighth the cost of Flash. >> Now, I wonder if can go back, I want to pick your technical brain for a minute. So you mentioned tape replacement, and then, as I recall the ascendancy of we can call them purpose built backup appliances, I think it's an IDC term or whatever, but we'll use that. A big part of the value proposition was plugging directly, looking like tape, so you didn't have to rip and replace your processes, and I remember Avomar was trying to convince the market that no, you have to change your processes, and people were like, conceptually that sounds good, but its too disruptive for me, so where were you guys on that curve? Do you look like tape, are you easy to pop in or? >> Proud to say we look nothing like tape. >> Okay, so that was a head wind for you early on, right? But it's really benefited you down the road, is that fair to say? >> If it was a head wind, it was a breeze, okay, and what I mean by that is, the technology we're referring to is VTL, Virtual Type Library, and in the very early days of the market, there were some legacy environments typically Fibersand type environments, where you had to make your disk look like tape so that the customer could transition, especially larger customers where, you know, change is harder, radical change is harder to make quickly. So VTL provided a sort of bridge, or transition technology over a period of time. We're through that phase of the market. >> Dave: But it was a band aide? >> It was very much a band aide. >> But you say it was a breeze, but Data Domaine got two thirds of the market, so, I mean... >> Yeah, but it wasn't because of their VTL. >> Dave: It wasn't. >> No, that was a result of there were still some Fiber environments out there, and they decided to cover that part of the market. We looked at the percentage of the market that we thought would need that, both in the early days but more, even more forward looking, you know, everything about our architecture is quite a bit more forward looking than the people we're competing against. And we realized that the investment it would take to do that, would eventually be wasted because it would go away, and heres why, if you look at what Veeam's software does with instant VM recovery and synthetic fulls and, sureback, and virtual lab, et cetera, when you make a disk look like tape, you lock yourself into the Fred Flinstone era of backup. In other words, you can't take advantage of any of the advanced features in that software, because tape couldn't support those features. And as far as the software knows, it thinks it's talking to a tape library, so it's doing silly things like saying fast forward, rewind, eject with disk. If you think about it, you can almost do a stand up set. >> Dave: Hey, your sequential... >> You know, picking on this, right. So what we said is, that's going to go away, it's very clear with what the software folks are doing, especially Veeam, that that's going to go away. Now, I realize Veeam recently added tape capability, but the reason for that is, not because its a primary backup media, it's because for customers that have, you know, infinite retention, or seven, eight, 10 year retention... >> Dave: They need an offsite tape option. >> They need an economic option. It's not that they like it, because we actually have a lot of conversations with customers, even with that longer term retention where they at least want to explore the economics of disk, but in some instances, even though they hate it, and they grin and bare it, they go with tape just purely economically. >> Right, so early days was, hey don't change anything about your software, keep the Fred Flinstone software and all your processes associated with that, and then, of course VM Ware changed everything. >> Marc: Right, and then graduate to the modern... >> Okay, and then the other big, sort of intern scenario, they used to argue about Dedupe rates, and I presume it's the work load and the nature of the data that determines that, not necessarily the technology, but maybe not, maybe there's some nuance on. >> It's a little bit of both. So a responsible deduplication vendor's going to ask the customer a number of questions about the make up and the nature of their data, okay, however, there's also a lot of aspects to which algorithm you use that are going to drive that. So, if you don't implement a very strong aggressive deduplication algorithm, your result is going to be lower, and we find in many of the software based implementations, and some of the appliance vendors, that they took shortcuts on the algorithms itself. Either because they were compute bound or you might be running it on a standard Windows server which is not optimized to run a really strong algorithm, and therefore where, we may say at 12 weeks of retention, you can get about 20 to one, they're getting six or seven to one, and in some cases they're recommending just put straight disk behind the software, well you end up with disk sprawl, because you're keeping all of this retention but you're not reducing the data enough, so you've got disk everywhere. >> Okay, so the quality of the data reduction algorithms matter, okay, and then the other arguments used to be inline or post process, Frank Luptin used "Oh that crappy post process..." >> Marc: I don't remember when he said that. >> Yeah, and weigh in on that. >> So, we kind of agree. Not that inline is better but that parallelization is better so we actually invented a third way called Adaptive Deduplication. Which basically, what that does is, it allows the chunks of data to land into our box first, and then we begin deduplicating, and replicating and parallel, right. So, we're doing it at the same time, but we're not doing it inline. And we monitor utilization of the system and we favor the backup window, so if think our deduplication is going to slow the back window down, we throttle back a bit. If we have plenty of resources, we crank away at the deduplication and replication. So we eliminated the potential drawbacks of post process, we eliminated the potential drawbacks of inline, and the biggest drawback of inline is that, when you go to recover a system and you think about Veeam's instant VM recovery, if you boot a virtual machine, we have that virtual machine in its entirety in a high speed cache, so it's up in seconds. So I was talking to a customer of ours at our booth who recovered an exchange server recently by booting it off of a Exagrid in about five minutes, right. If you tried to do that out of a dedupe, a device that only has inline deduplicative data, you're looking at hours to maybe even a day. Now you're CEO's not going to be too happy when they can't do email for a day, so I would recommend a high speed cache. >> Marc, Exagrid's been a partner with Veeam for a lot of this journey that Veeam's been on for the last 10 years. Here at the show, they've been talking about where the next 10 years are going, everything cloud, and expanding what they're doing, as you look forward, any announcements this week or as you look forward as a partnership, where do you see things growing? >> We don't have any specific announcements this week, I would refer folks to our website, we just recently announced our 5.0 release, it includes some pretty important things. One of the things it includes is, integration with Veeam's scale out backup repository, which dramatically simplifies the use of multiple Veeam repository's with Veeam's software. We also announced an offering for AWS we think that's appropriate for some customers, not all necessarily, where we can put a virtual appliance on Amazon, and in the cloud realm, there's no question that customers are going to continue to explore the cloud model for both efficiency, operational, expense versus capital, but there's going to be multiple cloud models, for example we partnered with a company, who's here, who you may have spoken to Offsite Data Sync. So if the customer doesn't want to do Amazon for some reason, then Offsite Data Sync will offer them the very same service with Exagrid technology and an operational expense model. And they've been a very good partner of ours as well. >> And the virtual appliance in AWS how does that work? You pop it in a COLO facility or? >> No, you literally, you load it into Amazon like you would any other Amazon machine instance, and it behaves just like a second data center. So you replicate to it, and it can store all of your offsite data, and then when you need it back, you can recover it provided bandwidth is adequate. >> So, I access the instance from the AWS marketplace, or? >> No, we actually provide it directly. >> Oh, okay. >> Through reseller network. >> Yeah, yeah, yeah, yeah. Okay, so I appreciate you by the way taking me down memory lane and sort of educating us on... >> Marc: Love talking about this stuff. >> Now, so, a lot of things we talked about are old news, to sort of set the context. Where are we today, what is the state of the market and the competitive differentiators that customers really care about? >> I think that we're at the state of the market where people are frustrated with a lot of legacy approaches, whether it's on the backup software side or the backup storage side. The licensing models are expensive, the vendors are gouging them, because they're trying to keep revenue, and they're worried about, you know, the players that are becoming the replacement players like Veeam, like Exagrid. So we're at point now where I see more activity of customers looking for alternatives to what they're running today than maybe in history of backup. You know, people always used to say, backup apps are very sticky, they're very hard to replace, well, look at what Veeam's been able to accomplish. Backup storage is very hard to replace, once it's installed. Well if you force a customer every three years to respend the money they already spent, plus more, you're creating a vent where that customers going to get frustrated and they're going to go out and look at alternatives. So I think we're at a point now where more so than ever, customers are looking for alternatives that stop the madness of backup spending, and stop the madness of backup performance degradation. >> Yeah, we had Dave Russel on yesterday and in his last magic quadrant, you probably read it, I think one of his strategic planning assumptions was 50% of the customers out there are going to replace or sunset their existing backup architecture in the next two years. I mean, that's a massive number, so, and obviously a huge opportunity for you and for Veeam. >> Yeah, I'm honored to be talking to Dave later today. >> Well Marc, listen, thanks very much for coming on theCube, it was really a pleasure. >> Thank you guys, it's been fun. >> Thank you. >> Thank you. >> Alright, keep it right there everybody, we'll be back with our next guest after this short break. (techno music)