 Okay, great, so we're back live, DaveValante, wikibon.org, with SiliconANGLE's continuous coverage. We're at theCUBE, at EMC World 2011. I'm here with my co-host, David Floyer. David, good to have you on. And we're here with Ralph Geordi, a customer of Atmos. We just had, we had Mike Feinberg on, who was one of the fathers of Atmos. You're a customer, obviously a recent customer because Atmos hasn't been around that long. So we're really dying to hear about your experiences at your organization. But first of all, if you wouldn't mind just sharing with our audience, tell us a little bit about yourself, your company, and your background. Yeah, thanks very much. So, good day for everybody. My name is Ralph Geordi. I'm a senior storage architect within Swisscom. Swisscom has eight tool eggs. One is a telecom provider within Switzerland. And the other one, we are also a service provider. So we host a lot of applications for our customers, our top 500 customers within our country. So 14 months ago, we had the task to evaluate a cheap storage solution. So we went out to the market. We did an RFQ and an RFA, et cetera. And then we ended up with two vendors at the end. Both were more or less the same price. But one was a lot ahead of the others in terms of green, in terms of power, in terms of rack space, and in terms also of the function. So then, we decided to go with EMC Atmos. So that's how we became the Atmos. So what was the, you said you needed cheap storage. Why was that so important in this particular area? It was important because our sister company, who is the telecom provider, needs to have cheap storage for their end users. So right now we have the tier one and the tier two and tier three storage, but this is much higher in price than an end user, customer, or consumer is willing to pay. So therefore for this one, we had to evaluate the cheap solution. So what the service was providing a storage for phone users or outcome users? What they do is they try to build applications where, for example, pictures or backup, BC online backups, stuff like this, where they can store their data into our cloud. Okay, and they don't want to pay anything for it? They pay something for it, yes, of course. So Ralph, you're a senior storage architect. You're not a senior storage cloud architect. You are now, I guess, but prior to that, I presume you have experience with traditional storage systems with a block or file. What's different in this world? What's changed? I would say the specialty here is, and that's the hard piece for me now, for my colleagues, is we have object-based storage. It's not file-based, it's object-based storage. So you have the chance to store metadata in it. You have the chance to apply policies, et cetera. So that's a big change in the mind of my colleagues because all they want to do is to store their files. And now with cloud storage, it's not that easy. That's the drawback right now. We have an API, we have a good solution, but we don't have the applications yet, the enablers, the connector, or how you want to name it. This is still missing right now. But the end users or the applications, they develop now against the API. So the ones, they made the mind shift already. They're absolutely happy with it. And the others, they have to do the mind shift. You're talking about a whole new perspective on adding business value because you can apply metadata principles, you can apply policies, and so forth. Can you, you said the ones that have done that and exploited the architecture more fully are really happy. Can you give us some examples so we can sort of visualize that? So, you mean in terms of policies? Yeah, well you said that there's a capability now with object storage that you didn't have before, and some people have taken advantage of it, certain applications, can you share what examples there might be? So for example, you can store a policy which means important. If this object is important or not, based on this, then you can apply a policy which means I need one local copy, I need a major copy, or I need the geo-parity. So that's one of the things. And also age-based policy, for example, I can say delete this object after 30 days, and this is something they will implement with the 2.0. So you have an age-based policy which automatically deletes the object, and that's totally new for them. So that automation and go ahead and salute it? So it's the method of being able to look at the method to find stuff, apply different policies, performance characteristics, et cetera. So you have to think about your architecture of your applications differently, I presume. Yes, it is. And that's the task of my colleagues who are the software architects that they have to rethink how they want to store the data. So we had an application which they're using a REST API and the backend, they're storing the data as an NFS application. So I came back and they said, why are you using this that way? So complicated, that's how we did it for the last couple of years. So I said, no, you have to change it. So I explained to them how it works, how the opportunity there are, and they're absolutely happy to change this now. Right, right. Because it's the right way to go. It's still a far way to go because, as I said it, they're not used to it, to an application API. But I hope, I hope, and I bet that within maybe a year or two, let's say, a REST driver will be part of every OS. Yeah, I think you're right. That's my wish. Yeah, I think you're right. But I think it's maybe a prediction as well. I called it a niche before, so just to poke Mike a little bit because we've talked about this, David, for a long time in the Wikibon community about the potential for this type of storage. It's enormous. It's just an exploding market. And the need to rewrite applications is one of the barriers to that growth right now, but it really is going to be exploding. What are you seeing there? Yeah, I agree with you that the opportunity in a number of different areas with storage now is what are the applications that can really exploit these new technologies? There are a number of new technologies coming in. One of them is Flash. We've talked about it before. Really, how do I architect new systems to take advantage of that? Development in completely different ways. Solve completely different problems with them and design something that is much more effective, much more intuitive to use, much easier for the end user. Why couldn't you use, let me just ask a really basic question here. Why couldn't you use traditional storage to do this? Why couldn't you use traditional file-based storage to do this? Too expensive? Well, you would have to put a database there. And once you put a database there, you're constrained by that database. So being able to separate out the metadata, put it somewhere else, being able to access that quickly, put that in a special place so you can access it quickly, just gives you automatically a whole much quicker way of being able to A, apply it, but B, more importantly, change it. And scale it. Scale it, change it. You can do all of that without having to touch any of the rest of the stuff. Is that right, Roth? Absolutely. Absolutely. So what's the current scale of the capacity that you're talking about? Can you talk about that on the users? How would you quantify that scale today? And where do you see it going? You mean the base installation? What do we have right now? Yes, yes. So the setup we have is, we have four different locations. So one, two are 10, 15 kilometers away, the others are 100 between each other. We built our internal network because we're also a network company. And from there, we set up also then the geoparity. So we forced the MC and we developed a very... So you asked you, you asked the MC for this? Absolutely, yeah, absolutely. So why was it so important for you? Because it's important for us to have a green solution and also to have a Jeep Jeep. And with geoparity, you can save a lot of storage because you only need 1.3 factory instead of two. And that's the first product who supports this. So we were talking to the MC a lot and say, hey, we want to use this and this and this in that way. And we did a lot of measure and also disaster, failure, and we took down the whole site and checked, are we still able to read the data, to write the data? What's the impact if a whole site is down? And then the original idea was to have first a local copy and then H-based or local mirror and then H-based replicated to geoparity. But with the experience we made and with our internal network, the latency was so low, then I said, no, I just go from day one with the geoparity. And that's how we work right now. So to answer your question right now, we have 720 terabyte draw. So we have four times six nodes installed and we're running now 1.41. And this obviously growing. The service is working, customers love it. Yep. And so if you looked out five years from now, it's just going to dwarf the capacity that you see today. And will this architecture take you there? If I can believe, and I hope I can believe my colleague's forecast, then it's going to be a couple of petabytes. A couple of petabytes. Yes. Right. And being able to respond that quickly with traditional storage and grow and scale, it just would have been so much of a challenge, too expensive, too complex. Yeah. I mean, we heard yesterday, so in the future, or the data explosion will be immense. And what they said yesterday is, if I remember correctly, it's roughly 90% is unstructured data. Right. And only 10%. And what we have right now is more or less, everything is structured data. And this is on tier one, this belongs to tier one, that's fine. But all for the rest, that's fine if you use it in the cloud. Basically, we heard you're right. I was listening to Tucci as well. He said that last year, 1.2 zettabytes were created in the digital world. A zettabyte is a billion trillion. Or maybe it's a trillion billion, either way. It's the same by the way, right? And then 35 zettabytes by the end of the decade, that's a 44X increase. That's... Did those numbers surprise you? Yeah. Yes and no. Because I just talked to my colleague who is running the whole operation business and he said he doubled, almost doubled the tier one within the first five months this year. So people must be buying it. Yeah, absolutely, absolutely. So it's crazy. So it's just going to take you to video as well? Have you got video up already or is that a whole new can of worms? No, also, we have a customer who will upload movies into it. Right. We have another internal customer who will also upload video clips into it. Right. These are the two applications which they develop now against the API. Yeah. And once I have a good, let's say the cloud drive or a good enabler for this, then it will rock. I'm sure, 100% sure. So Rob, this is very interesting. We don't talk to tons of guys do an object-based storage because they're not everywhere, right? But they're starting to become more and more prominent and more people are interested in this technology. What advice would you give your fellow storage colleagues that are thinking about moving to object storage and cloud storage? Maybe some things that you would have done differently if you had to do it over again. Share with your peers what you would advise them. So, good question, sir. I would say, or for me, number one is security. It's absolutely key. So I want to go, I want to sleep well. So for me, everything has to be secured. That's the most important thing. Especially in Switzerland. Especially in Switzerland. So yeah, we have the reputation of save money. So I hope we also will have the reputation of safe data. So again, that's key number one. So separate your traffic from the internal network versus the external network. That's key number one. Number two is do a good planning before how you want to set up it. Don't jump into it too fast. Think about, go one step back. Think about, rethink about how you would deploy it. And I'm proud to be able to say that I would build it the same again, in the same way. Absolutely. I'm standing, Ralph Geordi from Swisscom. Thank you very much for coming on theCUBE and sharing your experiences, sharing your knowledge. theCUBE is all about sharing knowledge and we've got hundreds of thousands of people out there watching this week. So thank you very much for coming on. Thanks.