 Live from New York, it's theCUBE, covering Riverbed Disrupt. Brought to you by Riverbed. Oh, here are your hosts, Dave Vellante and Stu Miniman. Welcome back to Riverbed Disrupt. This is theCUBE, the worldwide leader in live tech coverage. Youssef Kaledius here is the corporate vice president of Azure Networking at Microsoft. Youssef, thanks very much for coming. Thank you for having me. You're welcome. Keynote this morning, talking about Microsoft and Azure and the global network that you have built and are building. It's like the universe, it's constantly expanding. So, again, welcome and give us the update on Azure and your role. Well, we have been on this cloud journey now for a number of years. Many years ago, we started this thing called Azure. The cloud was at the beginning, if you will, just a buzzword, now it's real. So, adoption is happening like crazy, to be honest. So, everybody's moving to the cloud. Now we're experiencing amazing growth in the journey to the cloud at the moment. And as part of building the cloud, we had to build a huge footprint around the globe. It's a hyperscale system. We have 34 regions across the globe. We have a global wide-aided network that connects all these regions. And we are basically giving what customers want. We are close to what the customers are and we give them all the software that they need to run on that platform. Now, as an outside observer, and don't hate me for saying this, but you can correct me if I'm wrong. It felt like when Azure first came out, it was like, okay, we're going to do the cloud too. And now it feels like cloud is a way of life at Microsoft. Is that a fair observation? Well, I'll give you a small story. Actually, one of the few people who started Azure. That was 2006. And it was before, frankly, anybody used the term cloud, including us. We started by building something called the platform for the web. And we had a code name called Red Dog and what have you. So, what I'm trying to tell you is we started this journey way, way early on and we knew what we wanted to do. We wanted to build a very large-scale system, highly secure and multi-tenant that can run anything. That was the vision, actually, from 2006. Now, I grant you took us a while. We had a beta in 2008, then a GA a few years later and so forth, but we are all in. We are all in. It's one platform that runs everything the company itself has and it's open to a large, large set of customer, everybody else. So, Yusuf, when we talk to customers, end user practitioners, they've got their on-prem stuff. They're using SaaS, lots, Office 365, almost everybody I talk to is using it and there's public cloud. Can you talk about the importance of networking, kind of your space and how that ties together and especially what brings you to Riverbed? Definitely. So, it's really straightforward. You can't have a cloud without having a network. It's by definition. The cloud is out there. So, what's a cloud anyway? It's a large distributed set of data centers, capacity all over the globe. Between you or your customer or your employee, you have to reach the cloud somehow and how do you reach it? Through the network. So, when I say through the network, you start first with your client. What's a client? I see you have a mobile phone. You have a mobile phone. You start with the phone. You start with the laptop. You start with your device. You connect through your local service provider on ISP, wireless, what have you. Ultimately, you have to hit a network somewhere to get you to the cloud. So, the network is crucial to making a virtual cloud. And it can be quite complicated, by the way. There's infrastructure that starts from fiber in the ground, subsea cables, IP networks, peering, transit, the buzzwords are plenty. Net-net, without a first-class cloud, you cannot have a first-class cloud. Without a first-class network, you cannot have a first-class cloud. So, it's a couple of stats. 34 regions, 100 data centers. You're basically building out a global WAN. 1.5 million miles of fiber. You connect to 3,500 or more ISPs. You own subsea cables. Okay, so you're essentially building this global wide area network. I want to talk about that a little bit. And what role does RiverBit play? What's the partnership about? Maybe talk about that a little bit. So, it is very important to meet customers where they are today, in their needs. And customers need technologies, such as what RiverBit provides and others, to connect their applications to the cloud. From acceleration to connectivity, there's a wide spectrum of partners that are needed to make the story complete. And as you probably know, we in Microsoft have always been partner-friendly. We need the ecosystem. Customers want the ecosystem. So, we're very happy to partner with RiverBit, like many others, to give customers what they need for the connectivity solutions. So, what's the nature of the partnership? Is it a deep integration? Is it a go-to-market? Is it all of those things? So, Azure, for example, has something called a marketplace where you can go in there and find virtual machines, pre-tested systems, a way you can transact, and so forth. And we have most of RiverBit's solution already in the marketplace, and we're working to add more to them as well. So, we do that very well. We also have go-to-market activity as well with them and our partners. So, it's enabling RiverBit as a service, essentially? Most definitely. And NetNet for us is giving what customers want. And customers want what RiverBit has. I think we can talk about security a little bit. We don't usually think about sub-C cables on theCUBE, but you own sub-C cables. You obviously, there's physical security associated with that. You own all these millions of miles of fiber, and then there's obviously cyber security. Can you just give us the high level? What can you tell us about security? What should we know as observers? In this business, security, multi-tenancy, using the same infrastructure among multiple customers is a given, it's by definition, it's a cloud. So, from day one, we designed the system to be highly virtualized, both for your virtual machines to be separated and secure from each other, and for the network to be virtualized and also separate from each other. So, I can have company A, company B, who may be competitors running over the same network, because the networks are virtualized, and each one sees their own network, if you will. Ultimately, it's a slew of technologies you need to have in the platform to enable people to run a platform like this from a security perspective. Isolation, separation, DDoS prevention, et cetera. And we have all of these things. In addition, to go back to your previous question, we have partnerships with the security vendors, all the brand names you can think of, such that we can utilize them, including riverbed technologies, so customers can put the solution together to meet their corporate needs for security and isolation. So, it's a multifaceted multiple parts to put together a solution to go to market with. You described Azure, you sort of described the stack. So, the plumbing, the storage, the compute, the networking, and the security, and then the platform services, and then the tools, the development tools on top of that. And I guess the apps fit on top of that, right? So, really that's the fourth layer of the stack, simplifying it, of course. And then, you said, okay, we now connect as well to on-prem, and then we have this hybrid approach. So, Nirvana is same, same. Where are we in terms of achieving that level of interoperability and consistency? Of course. The market is in transition, as you might expect. There will always be things on-premises that are different from the cloud. By definition, the older staff, legacy, and so forth, all the very specialized systems. There will always be some, you have to do on-prem, if you want to do it very differently, if you want, very specialized. If you want to put a mainframe on-premises, et cetera. On the other hand, many things are becoming available on-prem and in the cloud. From a manageability perspective, APIs, identity, virtual machine management, and so forth. So, you're going to see foundational pieces that are very similar between on-premises in the cloud, where the cloud gives you hyperscale, a huge scale, and distributes geographies around the globe, and then you can put stuff on-premises in your data center, in your colos, whatever the case may be, to meet other needs. Isolation, geo-political considerations, et cetera, or latency for that matter. So, it's an evolving picture, and there will always be different things in different parts, but for a long time, we believe that the hybrid model will dominate. So, you're talking about scale, and I wonder if we could ask you about the economics of scale. Microsoft taught us the power of volume economics with software, and its ascendancy in the 80s and 90s. Is there a similar analogy with cloud at volume? You're packaging all these services, you're automating to a high degree, you're eliminating all this heavy lifting and human IT labor, and super high degrees of efficiency. What are the implications on marginal economics and ultimately pricing for the customer? It has a few implications. One is that the market will eventually evolve into a few large providers, that's what we believe. It's just the economics of the situation is such that it's hard to have many, many cloud providers, to be honest. You're going to have many others adding value, being integrators and so forth, but when it comes to the basic infrastructure, I believe the number will be very small going forward. And again, the reason is you mentioned subsea cables, these are not cheap. You mentioned 34 regions and more than 100 data centers, these are not cheap, et cetera. So this is a scale beyond which it's going to be very hard that otherwise you have many players in the space. For the customers, they will actually gain many advantages that we are obtaining by our scale. Frankly, when we buy our servers, we buy them by the truckload, literally. When we buy capacity, we buy them in the terrible, bits per second, numbers. And as such, you end up with economies of scale that we are invariably pass through to our customers. So in both ways, the industry is going to become, I believe, a few big providers for the infrastructure and invariably the price reductions if you will go to the customer. And you always hear people talk about the race to zero, but it's not a race to zero, is it? I mean, it can be a highly profitable business. Definitely, and please note, what customers really want the cloud for is because of agility more than anything else. Don't get me wrong, the cost structure is good too, but the agility of standing up a new service, connecting it, and so forth, sometimes literally within days versus years or months, that's worth gold. People would pay more for that too. Yeah, Yusuf, I think Microsoft is a great story on hybrid, we talked about the various profiles from a networking standpoint. Curious what you're seeing from customers when it comes to things like cloud data migration and cloud bursting. We've watched Amazon with their snowball, cloud bursting, something we've batted back and forth for the last five years or so at least, but what do you see from a customer base? Well, the need for connectivity is increasing, both from the actual consumer and the customer to the cloud and within the cloud if you want. So you're going to see the numbers always increasing in terms of bandwidth and the like. The scenarios we are seeing for cloud usage are actually very, including for recovery, backup, disaster recovery and the like, which requires a lot of data to be transferred, both ways, by the way. Sometimes you want to go to the cloud, sometimes you want to exit to the cloud. There will always be niche corners where you need very, very high bandwidth, where the fastest way is a pile of tapes. You know the old joke. Sure, CTAM has the fastest. Exactly, again the 747, 424 tapes, that's the highest bandwidth solution, the very low latency. So there will always be these niche solutions which we also support, by the way. But net-net connectivity and ubiquitous connectivity, both for the clients and for the backend, is the way it's going to happen going forward. How about, I want to go back to the hybrid for a minute. We often say in theCUBE that cloud is not just about where you put applications and data, but it's about the operating model. Sure. So let's talk about what that means to IT executives and business people. What does it mean to have a cloud operating model and how can I as a customer achieve that? That's a deep question. I'll touch on some of it by relating our own experience in building the cloud and operating the cloud. In the early days, we used to, many of us come from enterprise software background, where we used to spend 18 months, 24 months building a new software release. And we had time. We'd spend six months designing, a year writing code, a year testing, and you cut the CD and ship it out and you have a party and so forth. Things were slower, very deliberate, but we could not really meet customer needs if you want. Now things are much faster. If things are agile, from the way we develop the software, the way we operate the software. I believe some of these aspects will reflect onto IT as we speak. And actually they're going to be a good thing for IT because it will elevate IT from being, if you will, a cost center or organization that's trying to run the business to become core to the business. And it will help IT spend less time doing the plumbing piece. We're the plumbers, we actually build all the lots of the low level stuff and the infrastructure. And they can move up the stack and add value and be very agile to meet their business needs. The implication of this thing on how IT would operate include the need to embrace some new technologies while still retaining the ability to connect back to on-prem, do the hybrid stuff so you can have both sides, if you want. So transformation is the name of the game here and the end game is really agility. Does that need for speed have, even if it's short term, potentially midterm effects and maybe even long term effects on quality, how do organizations, yourselves included, square that circle? You'd be surprised, I mean you can take a piece of software test it forever and then ship it and have a long time before the next version and you're going to have lots of bugs in between as opposed to shipping often and doing the right testing methodology where you can actually recover very, very quickly. The basic techniques we now use, we roll forward, we roll backward, we test and that's more before we go big and so forth. So what I'm trying to tell you is agility is actually the main tool we use to get higher quality now as opposed to time in the past, which was let's go bake the bits. We used to bake them in the lab and then still find the bugs. And you have, I presume you have proof points of that. I mean you have the data. The cloud as we know it today will not operate otherwise. So it's more, it's obviously faster, more agile, it's higher quality. Yep, it's very true actually, I strongly believe that. And more secure by the way. I was going to ask you, is it more secure? How many features scrolled throughout during your keynote there in the last 12 months? That's actually just a snapshot of what we have done. But actually it is more secure because we apply the security knowledge we have and the tooling and the data to the global cloud. We have a global view of all the potential attacks and so forth and we dedicate teams for the security of the infrastructure. Sometimes frankly when we used to give it to customers some customers were amazing, others were not as good. So if you will extract all these things and put it in one place now in the cloud itself. Security used to be the number one reason why people didn't go to the cloud, particularly in financial services and healthcare. And then they turned around and realized, wow, they're all actually used for how. You can actually argue that security is one of the main reasons why you want to move to the cloud. Yeah, so why? Within the reason of course of your political boundaries and so forth. Yeah, compliance and so forth. So why don't people move to the cloud, why? I think they are moving to the cloud. But do note, customers also have some costs on premises. And we as human organizations sometimes takes us time to switch the earlier question, the operating model and the like. But the movement is happening. But if you asked me a few years ago, it was mostly test, dev and test and proof of concept. Now the movement is really happening. It's inevitable and not just because you're biased because you're in the Azure. I am biased, but we are seeing the numbers speak for themselves. But Microsoft believes that now. That's what as an observer, I... I can give you some numbers, you know. We sign up 100,000 subscriptions a month now. 100,000 subscriptions a month. And the numbers keep growing. You know, I remind you of the footprint. The reason we have 34 regions, which is more than what Google and Amazon has combined is a testament of the demand we are seeing. The reason we have this huge network you mentioned earlier is because we need it to connect all this infrastructure for the cause of the demand. The demand is there. Massive scale, agility, you know, huge investments, obviously. Yousef, thanks very much for coming on theCUBE. Appreciate it. Thank you guys, I appreciate your time as well. You're welcome. Keep it right there, everybody. We'll be back with our next guest. Right after this, we're live from Disrupt in New York City. Right back. My name is Dave Vellante and...