Andres Rodriguez, the founder and CEO of cloud storage startup Nasuni, sits down with SearchCloudComputing.com's Carl Brooks to discuss where cloud storage, and cloud computing in general, is headed.
Andres and Carl also discuss:
- How Nasuni melds with cloud storage services like Amazon S3
- The potential for change in cloud architectures
- Any difference between cloud storage and online storage?
- Cloud computing innovation from Google and Amazon
- Experimenting with cloud computing techniques
- Who besides Microsoft wants to beat Google?
- Economies of scale and the value of cloud
Read the full transcript from this video below:
Cloud storage startup CEO analyzes the future of cloud
Carl Brooks: I'm here with Andres Rodriguez, the CEO of Nasuni, a Cloud software start-up, based here in Massachusetts.
Andres Rodriguez was CTO of the New York Times and an entrepreneur of several start-ups including Archivas, which is a content management solution that was subsequently bought by Hitachi in 2006 and been described as sort of a precursor to Cloud storage and some of the new things we're talking about today.
Why are you back in the game? What's interesting about the market today? Why come back with Nasuni, which is, you have a product called the Filer, that allows people to run in their [SAN] and become a storage and back up solution and sort of seamlessly blend out into Cloud storage services like Amazon S3 and so forth. So why come back in with this now?
Andres Rodriguez: Thanks, Carl. First of all, thanks for having us in the show.
Because what I did before was an archetype to the way that Cloud storage has turned out to be, the Cloud stores that we're talking about is things like Amazon S3, things like RackSpace clouds files, Amazon with Asher. We saw very early on the potential for these architectures to really change the way that IT does their day-to-day tasks around storage and, specifically, the change was around how to protect the data.
This isn't so much a play around just plain storage. You can get storage in hard drives and forget about it and they're very cheap and all this stuff. But what's really painful in IT is how do you protect arrays, how do you protect file servers and these architectures of the big players like Amazon; they had to contend with incredible growth. Growth that is so much faster than anything we've ever seen on even the largest enterprise accounts that things like backup had to be re-thought from scratch. They developed systems that didn't need the traditional backup systems to keep data protected. They made compromises.
What we saw an opportunity to do at Nasuni was to correct some of those compromises and be able to deliver that model for data protection to an end user who's an IT person who's sitting there and they've got a file server and they're having a lot of pain around, say, backup.
What this is, is a transformation of how you deliver very familiar IT services in a way that is far less painful and more reliable to the IT administrator. That's never happened. It's very rare, specially that a retailer, a book seller like Amazon or the web properties, all of the sudden have to acknowledge it. I can tell you, I think what they have is superior when it comes to scale and data protection and availability, than anything that we've seen come out of the large storage OEMs, for the last 5-10 years. And that's what's exciting. What's exciting is to connect that much potential with the traditional way in which we do storage and we'll continue to do storage for a long time. In our case, it happens to be file servers. Other companies are going after storage array technology but, to really tap into those repositories.
Carl Brooks: Clearly Amazon thinks that it is doing something with storage well enough to be able to sell it to other people. Practically you can't open a newspaper today, let alone turn on the internet without hearing the term Cloud computing. How is this fundamentally all that different from online storage or any way you might approach like DR or just finding a way to store your stuff safely? People have been hosting and storing stuff both off and online and off-site for ages.
Andres Rodriguez: That's a great question.
Carl Brooks: What is Amazon's or whomever's storage model different?
Andres Rodriguez: It's really two things: Its cost and its size. Amazon and Google and, to some extent, Microsoft, they've already built the largest data centers in the world because their own businesses needed that. What they're doing now is, they're opening up that infrastructure to the public for the first time and so, a mid-size business, even an enterprise business, now has access to some of the most reliable infrastructure that is well beyond anything they could do themselves.
For instance, think about the way that you protect data with SecondSight. What if, instead of just giving you a SecondSight for data protection, I can give you three copies of your data, in one data center and then that replicated across three or four data centers around the world and have that all be policy based, all ran behind the scenes for you and offered to you as a service?
All of the sudden, that's a model of reliability that's much better and much simpler for you to access than anything that you've been able to do in the past. That's what's exciting about it. It's access to that level of quality.
Carl Brooks: So the innovative things that Amazon or Microsoft or Google is doing are now, they're able to find a way to let people into that. Things they would not be able to do themselves. That's what makes cloud computing different or...?
Andres Rodriguez: That is one of the big differences. Remember, these people had to develop, the Googles, the Amazons, they had to develop some of the best practices and best software. Best systems for doing storage, for doing compute, for doing provisioning, for keeping systems 100% available all the time, to all their services. You're tapping straight into that. That is well beyond what most businesses do and definitely what most businesses can do at the cost that those providers are offering those services. That's where the transformation is.
Carl Brooks: The thing is everybody's got computers, for the most part. People running IT operations are usually, probably, drowning in servers, or storage or what not. They've got lots of these resources around. Is there some shift to how people are thinking about what they want to get? Are IT pros going to have to do this in their own shop? Are they going to have to turn themselves into little mini-Amazons and mini-Googles?
Andres Rodriguez: It's a great question. There's this dichotomy between public Cloud and private Cloud. A lot of people are thinking they can copy the systems and the best practices out of the big players and, say, do storage clusters for storage. The same with an Amazon, has built the largest storage cluster in the world and do fertilization farms where you can have actually elastic computing. The same with, Amazon has done with EZ2.
My take on that is that, I think there's going to be a time for doing that and a scale for doing that. But I really think that is just a temporary solution that makes both the large enterprise customers and the large OEM players feel safe. Feel safe because the OEM players can continue to sell gear and software which is something they love to do and the enterprise players feel like the data is still under their control.
That may take 5, 10 years to develop. There's a lot of opportunity there for people but, I think that, ultimately the end goal for a lot of this stuff is massive, ten 90,000 server data centers cooled by rivers, that are being run at very high levels of reliability. Huge economies of scales for how you buy the equipment and very sophisticated software for how you provision and de-provision the systems to make sure your reach these very high levels of service at very low cost.
Only the biggest players can win. I think, actually, what's going to happen in the next two years, we're going to see a lot of the players that we used to think of as being large players in Cloud drop out because it's a game for giants. This is not a game for just big companies; it's a game for giant companies that have already a lot of their own business riding on this big data center.
Carl Brooks: Well, conversely, though, there are many, many enterprises that are not involved in IT that operate or come close to operating on a scale that Microsoft or Google operate on. I talked to a large automaker and, it was at a show and I talked to one of the fellows who does some of the design in their forward looking things and said, "We're experimenting with Cloud computing techniques. We're probably looking at 18 months to 36 months out to do really large deployments" and I said, "But what about this new data center Google's building out in Utah or some God forsaking place" and he said, "Well, it's a nice center but I've got three of those." They're really my concern.It seems like there maybe lessons to learn from Google but we can't really... Who's going to beat Google? Who's got that many smart people?
Andres Rodriguez: And the question is: who wants to beat Google? In other words, for Microsoft to want to beat Google, it makes sense. They're both competing in the same space. They both need scale.
But for an automaker to beat Google, it may not make sense, even if they can pay their way into it. And, definitely for the mid-size business, for the IT Administrator that is just trying to protect the file servers or is just trying to provision more storage into their arrays, it doesn't make any sense to build their own little mini-Cloud. It's much better to just tap straight on. The idea here is, if I want a terabyte, I can get a terabyte in a minute, in a second. I don't have to go and get disks. I don't have to build my own system for it.
Carl Brooks: Or I can go to Best Buy. I'd probably be back in about 35 minutes for the terabyte. I mean, that's the trick, isn't it? We have convenience on one side that we've built this hardware race. It's amazing. I can buy a router at Best Buy. 10 years ago, I would have had to write myself. All the storage and all the stuff is out there but then, go to the online world, the availability worlds and is yet easier to get that stuff. Just yet that much more easier.
Now, one question that always comes up and every single conversation I have is security and security and security and security. And one of the things people cite is impudence is a barrier to using large-scale cloud computing, as you say, we probably will be doing in the future is the idea that data can't be out of your control.
Just to give you an idea of what this means to a lot of people. I talk to an enterprise IT Architect and he said, "We're never using cloud. Not in a million years. My CIO, he takes care of us, he takes care of all my users and he's keeping everything safe." And he spend 20 minutes telling me about his new hosting at AT&T, that they've got the business data on. So this clearly is a little bit of a mismatch.
Now, I know that, you guys have done some interesting things in terms of both encryption and keeping data safe in a public cloud environment, which is of, I don't know, is it of dubious safety? Are there ways that you can address this safely?
Is there a way that an enterprise can actually go and use like S3, not by buying your products? Is there a policy way to address this inherent insecurity?
Andres Rodriguez: The root cause of insecurity in the cloud is that people feel that because is a [multi-tenancy] environment, because many people are sharing in the resources of the Cloud, there is risk that someone else might accidentally or maliciously look at my data.
Now, in a cloud, in a compute environment, if you're doing computation on that data, that data then needs to be visible to the provider in the cloud. And that is a huge potential security risk and I think it will slow down cloud computing for a while.
In the storage space, the benefit is that you can have all the data, so all of our customers' data is encrypted at their perimeter, in their data center, with their own encryption keys. When you're encrypting data, using something like AS256, the ability for either the provider, Nasuni or anyone outside the perimeter, anyone that doesn't have the encryption keys to actually be able to decipher that data, is close to an impossibility. This is where it would be much easier to just drive a truck through the wall of your data center than it is to penetrate AS256. Top Secret materials are done that way. Banking is done that way. There's so much stuff that's already riding on data encryption so I really think that you have to think carefully about.
The promise to our customers is we make your data as secure in the cloud as it is in your data center. If someone comes in and steals your encryption keys in your data center, now they have access to your data but that means you had a breach of security in your data center. No one can come in the back door through the cloud and get to that data. And I think that you always have to weigh the benefits versus the risks. In my company, we give Salesforce.com all our customer data happily everyday because there's so much value to the service that they bring in. And I think that a lot of cloud storage is also like that.
We are locking on the data, we're offering that level of security but, at the same time, we're delivering tons of services with it, tons of benefits.
Carl Brooks: Well, that brings me to the last I'd like to ask you about and that's value or economic imperative, if you want to use $6 words.
We've seen companies, as we said earlier, like Google, like Amazon, like Microsoft to a lesser extent, really take this idea of economies of scale and push it way out there. And they've managed to prove that, Amazon itself has managed to prove that, you can run a data center in such way you can turn around and sell computing storage for pennies, literally pennies to anybody they want to, for all commerce, essentially for all time. Is that essentially what is going to push this ship forward? Is this an economic necessity, an economic reality?
Convenience is one thing but there's many, many businesses and many, many IT operations that aren't run by convenience but by a different set of needs or a set policy. Quick, easy, cheap. Does that have to be the third indicator here?
Andres Rodriguez: Let's go back to the Salesforce example. Why was Salesforce.com successful?
Carl Brooks: Because Cyborg was awful?
Andres Rodriguez: Yes. Exactly. That's exactly why. Because what was there before made sense, if you're willing to make a huge investment and then a huge investment, not just in setting it up but in running it, overtime.
Salesforce came in and, it wasn't that they were cheaper. It was that they were so convenient, they could be broken down into small little pieces and you could bite in $100 at a time. By the time you were running a 500 person organization, Salesforce's says you were paying more than you were paying with traditional CRM systems.
The Cloud is not different from that in that, what we see from our customers is, the number one reason they buy the product, is simplicity. They just want a simplification. There is pressure from data growth. That puts pressure everywhere. Primary arrays get overloaded and my entire backup infrastructure and reputation infrastructure is overwhelmed by that. So just help me take some of this pressure off in a really simple way. And I can bite into it a terabyte at a time.
That's very appealing. I don't have to go out and buy a two, four terabyte array. I can just do it a terabyte at a time. So very much like the Salesforce model, this is a model of delivering functionality that doesn't appear to be much different or revolutionary. It's just a heck of a lot simpler and I can bite into it in small enough chunks to make the transformation over a certain period of time.
I think that's what the cloud is about. Is convenience, is small bite-sizes that you bite into on demand, when you need it, it's there.
Carl Brooks: And so that's going to drive adoption or, at least, that's going to facilitate adoption which is going to be driven this overall necessity to find the same kind of economies of scale that big innovators providers have.
Andres Rodriguez: You want to tap into the scale without having to build that infrastructure yourself. Right? That's the key. It's just like Salesforce. You want to tap into the functionality without deploying your own CRM system. That's the potential here.
Carl Brooks: And I think it's interesting to note that Salesforce.com, the core product, has now gotten to the point where there's a whole ecosystem of smaller providers around it that overlay on top of it and provide services for your Salesforce.
They've gradually increased in complexity and range of features to the point where it sort of builds on itself.
So there you have it. Thank you very much for watching.
I'm here with Andres Rodriguez of Nasuni and I'm Carl Brooks here in Boston for Cloud Cover TV.