The next era in consolidation
Date: Fri, 08/24/2012 - 14:48
Consolidation continues to be a compelling initiative, driven by the increasing prevalence of server virtualization and virtual desktop infrastructures (VDI). These powerful technologies have introduced ways to reduce cost, and complexity while offering IT management higher levels of control and security
Carolyn Crandall, Riverbed's VP of Worldwide Marketing will introduce to you an evolutionary new concept in IT consolidation. Much as VDI has enabled consolidation of desktop infrastructure to the data center, the introduction of an innovative technology called edge Virtual Server Infrastructure (eVSI) will now enable organizations to further centralize their servers and storage into the data center. Primary use cases such as complete consolidation, rapid service deployment (VM Projection), and lower TCO are all drivers for the interest and adoption of this new technology.
Carolyn will also share challenges companies face and how visibility into performance could be impacted by consolidation, along with ways to overcome issues related to a distributed enterprise: reliability, latency over distance, and traffic bottlenecks. She will show how businesses can increase their productivity, efficiency, control, and resilience.
Guest speaker presentation by Carolyn Crandall, VP Global Marketing, Riverbed Technology, at NetEvents EMEA Press Summit, Garmisch, Germany
Hi. Thank for you that warm and fun introduction. He told me last night he was going to give me a tough time today, try to give me some questions I couldn’t answer later on in the Q&A. I will try to be ready.
So one of the things I want to talk to you about is -- I was asked to do the keynote today - is to share something that is a little bit more visionary. As I thought about the topics, I obviously wanted to do something that related to an area that Riverbed was focused on, but also to introduce a new technology to you and not talk about something that you may have heard from other people before.
So the topic I centred on was really the next wave of consolidation. So on that front, there's two areas of consolidation that we're likely familiar with today. So, for some, I apologise, this may seem a little basic at first but I promise it will get more interesting as we get to the third area. But the ones who are more familiar with this virtualising, our desktop infrastructure and also virtualising our servers, I'll talk a little bit about some of the challenges as people have gone out and virtualised. They've realised there's been some compromises and there are some things that are not quite as complete as they would have wanted it to be as they started to consolidate. Then what I'll do is go and talk about the new technology, which is the virtual server infrastructure, which is something that has not been done before. This really affects all aspects, so whether you're in the branch office, whether you're in the data centre, whether you're an individual user or part of the staff. It actually makes it a very fascinating topic as we go in and speak, both with the CIOs and the people that manage the technology within our customers.
Probably the first thing on server virtualisation, is it happening? Is it even relevant?
Do people care? From the Gartner study that was done what we saw is that 40% of companies have already started to virtualise and they expect that number to double.
So as we look at the opportunity here for companies to better manage and consolidate their information, it's been widely adopted.
Let's first talk about what it is. I know there's varying levels of knowledge within the audience today, but if we take a typical server environment where you have your A, B, C servers, and, basically, as you do server virtualisation, you consolidate them all on to a machine in itself. So one of the things that that allows people to do is lower their costs; they don't have to manage as many servers. It allows them to have more manageability and control. It also allows them to underpin most clouds which again, as people are looking at cloud initiatives and other things, this helps make it a reality.
The other side of things, in addition to why people virtualise, VDI, which is your virtual desktop infrastructure, is being deployed at about 21%. So again it’s a little bit less than server virtualisation, but it's something that is being adopted more and more today. So very similar to -- here's the before and I'll give you an after situation -- where you have your different desktop devices, your A. B and C. As you do VDI, you basically are taking and pulling them into the server and projecting them back out.
So a lot of people have liked that for rapid deployment of things, for control of their desktop infrastructure. Again, it's fairly simple. It's an architectural design that a company would choose to do if they wanted to have more control of their desktops and being able to rapidly deploy things.
The other thing that that's given them is the ability, whether they decide they want to change where they're projecting the image of things. So let's say a user moves around; they may go from being in this environment to now needing to have it projected out into a different area, or potentially they may decide that they want to go from a laptop over to a tablet type of a device. With the VDI infrastructure, it's given a lot more flexibility to be able to deploy quickly and easily on multiple types of devices.
Just in quick review, server virtualisation allows you to have less physical servers, all on one machine, and VDI allows you to project the desktops from the data centres.
So, in the end, what they're looking for is happy office users and a happy IT staff because they get that control around the consolidation that they've been looking for.
As I mentioned, I will share a little bit about what are some of the problems or challenges that people have had when they’ve done this. One of the big challenges that they've seen in data centre consolidation is loss of visibility. As people go and they consolidate those servers on to these virtual machines, they lose a lot of the typical traffic patterning that they could see before. So when they want to understand where are the problems in the network, they can't necessarily understand that.
Secondly, if things are up in the cloud, and the cloud is able to move them around, they have less control, if they have performance issues, to be able to recover that control. Today I'm not going to specifically spend time on the network management tools that have been created, but there are tools now that make that easier for people to understand what's happening, even in a cloud environment and, therefore, again fuelling the adoption of consolidation and virtualisation.
The second one is really around the consolidation of the edge servers. You've been able to put virtual servers all on to one machine, you've been able to virtualise your desktops, but your servers are still sitting out in the branch environment. Many companies or government agencies that have goals towards 100% consolidation, have found themselves challenged by being able to get to that point because they have to leave servers within the branch. Yes, and the unhappy network team because they can't accomplish their 100% consolidation goals.
Moving from the two, as I mentioned before, virtual servers and VDI, we're now introducing a new technology called virtual server infrastructure or some people often call it edge virtual server infrastructure to delineate that it at the edge. Again, very much like VDI. What does it do? It allows you to project just like VDI did. Instead of your desktops it's actually projecting your servers out into the branch office location. So a visual look on this. You have your actual, physical servers that are out in your branch offices. You'll be able to move them all into the core data centre and then project them back out.
One of the startling things we found with our customer base is that three-quarters of our customers were leaving servers in their branch office. We thought with WAN optimisation technology, why wouldn't everybody put things back into the data centre.
What is keeping people from consolidating the rest of their information?
So as we looked at it and talked about it to our customers, the immediate reaction is, wait, you can't do that. You can't boot across the WAN. If you try to boot across the WAN, the performance is going to be so bad that we'll be sitting there waiting and it will just time out and fail and it just won't work. So we need our servers to be able to boot across the WAN. We have custom applications. When you use those technologies like WAN optimisation to accelerate once you virtualise back into the data centre, I have this custom application that nobody had written the acceleration for.
So, therefore, I need to keep the server inside of my branch, otherwise that won't work.
The third thing is what if the WAN goes down? So what am I supposed to do if the WAN goes down and we need to keep our operations running?
These were the three major barriers that our customers presented in front of us that said this is great technology, but I can't use it. So what we've done is designed a technology that, if you look at the diagram here, a little bit more of a networking diagram for you, but the little blue here and the other piece, the blue box here are, what we call, our Granite technology. Granite is this virtual server infrastructure technology, as I was defining before. The way that this would be set up is you would have a box inside of the data centre and you would have either a second box in the branch or you could have the Granite technology running on your WAN acceleration box or Steelhead appliance today. The great thing about this is that it allows the people to take their servers and their applications, virtualise all their services and servers and to consolidate them back into the data centre. Basically you're consolidating your VMs and then you're projecting them back out across the WAN.
Then when the customer goes to boot, it takes less than a minute for them to boot completely across the WAN. It's pretty amazing technology, when you think about that, because now you have an environment that you have complete control, 100% consolidation from the IT, but you also have those happy customers that are going and not feeling a change in performance.
Some of the other things that we have done here is this custom application piece, that we said before you could not consolidate. It's a different type of technology than what you've seen in WAN optimisation and other things. So it will now again be able to address the read-writes on that so that you can consolidate. A lot more technical detail behind what the products actually are, but from a very simplistic standpoint, it solves the issue of being able to boot across the WAN. It solves the read-write issues associated with custom applications and keeping that all local. Then also there is enough storage or cache on the device that in the event that the WAN goes down, operations will still stay live because it will continue to run off of the local image on the device.
Taking it from an idea of what it is in a new technology, the question is do customers want this. If you talk to a lot of people that run a data centre; if you ask them to draw a diagram of what they would want their network to look like, pull out the whiteboard, and they'd walk up and they'd put one dot on the whiteboard -- and maybe a little dot over here for business continuity and disaster recovery -- but it would be one dot.
They would want it because they would want the total control of their network. This allows them to have that complete consolidation and control.
The other difference is when they looked at VDI, they found that VDI didn't really save the money. It gave them control, but it didn't give them the savings. The difference here is that the customers that we worked with from a data perspective have realised over 50% savings from deploying this virtual server infrastructure.
So what I'll do is I'll run you through four case studies of beta customers that have used this and why they've chosen to stay with this technology. The other thing that they found is, unlike a few years ago where everybody only focused on cost savings, now people are more concerned with customers; the customers and users and the experience. They don't want to have a bad performance problem. So we've been able to address not only the control and the cost, but now also the performance with the customers.
Then the fourth case study is really around VM projection. That is great in environments where people need to be able to project things to a certain area, but then pull them back quickly. Or maybe they want to project a new image or a patch out to multiple branches at the same time. So instead of having somebody within each of the different branch offices have to do this, they can project it all from a core data centre environment, which ensures that everybody gets the same information at the same time and dramatically lowers the cost of that deployment.
Let's talk about the first one. This is a Mid-West law firm. Their initial investment was about $730,000 for server infrastructure. They invested about $326,000 on this virtual server infrastructure. In turn what they were able to do is completely eliminate their servers and their main benefit was being able to pull the storage back into the core data centre and leverage core storage that they already had there. So this is a hard ROI story where they were able to demonstrate, strictly on savings alone, the benefit of going over to this technology. What you'll see also here is there are big use cases and smaller company use cases where this technology makes sense. This was about 600 attorneys in about 10 branch offices.
This one is an entertainment company -- you might be able to guess, I can't say who they are -- but they were going into China. They had a lot of data privacy and security concerns. They did not want their information to be in China. They thought the risk of it not being handled correctly was too high. What we did is we worked with them and said, okay, it's an architectural design. How can we allow their people to get access to and interact with the data, but at the same time mitigate any risk or concern.
Again, in this case, they were able to deploy the virtual server infrastructure, or our Granite technology, so that they could project the servers, their employees could go in and interact with the data. Then, when they were done, they would just turn it off and then there would be no data left within China and that mitigated their concerns. So they were very happy with this from an ability to, one, consolidate back into a core data centre and then have the control around their information that they wanted.
This one's an interesting one too. Engineering and construction companies, as we know, are very mobile, are transient. They don't stay in one location permanently because they set up the building or construction and then they'll move on. This one they have, at any given time, around 300 job sites running and anywhere from 5 to 500 workers per job site. They needed something that was going to be scaleable for them and allow them to provision things out instantly and, in the event that they had to change a location or a place, that they could instantly redeploy that. So the VM projection piece of it was particularly interesting for them in this case because they were able to not bring local infrastructure on site and able to just use the Granite technology to set up and project their servers, have their employees again interact with the information and then move it to the next place as they needed to. Also if they wanted to communicate out to several of these branch locations all at the same time with updates, they were able to rapidly do that as well.
This one is a multi-billion dollar publisher. They were one of the companies that was trying to tip towards 100% as far as they could get. They actually put a mandate in as a company that said that we want to get at least to 75% consolidation. The problem was that custom application piece; they used something called extraneous IP and nobody had been able to accelerate it. Every time they tried to consolidate, they couldn't because their -- and customers or their end users were completely dissatisfied because it didn't work. They had a Mac environment, they had this custom application and they were very frustrated because they couldn't come up with a solution. So what this allowed them to do was to accelerate that application, as I mentioned before, and then meet their server consolidation goals. Then, ultimately, the organisation was very happy because they were again able to save about 50% per site in savings. So it was, again, a very interesting solution for them that they were excited to adopt.
The other thing that I'll say is that the beta customers -- we actually announced this product this quarter. The beta customers that we have, really none of them wanted to give the product back. They got it in and they used it and they said this is fantastic.
The thing that people need to recognise, this is an architectural change. Not everybody is going to want to go do this. We're going to have many people that will want to stay with traditional WAN optimisation and have a mix of consolidation and still branch office. There will be other people, for the reasons that I outlined, that will want to go to as close to 100% consolidation as they can. The nice thing about this is now they have a choice. They can consolidate as much as they want. They can start with the virtual server piece of it, they can go to VDI or they could go to this full virtual server infrastructure integration. Again now they have steps or tiers that they can go through that gives them maximum amount of flexibility.
Again this is a high level overview and I know that there's some fantastic questions coming my way now.
Guest speaker Interview & audience Q&A
Manek Dubash – Editorial Director, NetEvents
Great stuff so thank you very much for that. I have to say, the first question I have to ask is perhaps going back to basics and relates to some of the experiences I've had talking to end users on forums, data centre forums, those sorts of things. The first question when you ask them what do you think about WAN optimisation -- so let's get back to basics here -- is why is it so expensive? They really find that they can't afford it.
It's a great question. But it's interesting, when we look at our customers and the experience that they've had, IDC did a study and what they found out is that the small to medium-sized customers that we have get an average payback in about six to nine months. Although some people may feel it's expensive, if you look at that window for payback it's much better than many other IT technologies. Although it is admittedly an investment, I think that people see a very rapid return. The other thing that we see is, we very often have what we call a proof of concept or an At Eval program in the company. More times than not we can't get those eval boxes back because once they experience it, they don't want to go back to what it felt like before from a performance standpoint. So we feel that our ROI story on that is very strong with our technology.
Okay. You said in your presentation that you can cover off an outage of the wide area network. I think this is probably the biggest nightmare of any data centre manager or network manager, is the network goes down. Of course, if you're in a cloud deployment, if the network goes down, it's totally network dependent. You said you're covering that off. How do you do that?
Yes. The main thing is, it's within the cache of the box. So there is a certain amount of memory of storage that is on the box. If the WAN does go down, it has that memory or information stored there which will hold people over for normal WAN outage periods of time..
It just depends. It depends upon how much information or data that they have and how frequently that information changes. But what we found is that this has been an acceptable amount of time until they can go back in and restore.
So basically it's a big cache; it's a lot of disk. Doesn't that imply that effectively you're storing a lot of data; it sounds like a server or certainly a storage sub-system.
How is that different from what they already have?
Well I think it's the other things that the box does. It removes the functionality that you would normally have to have in a separate server environment. So one is the reduction of cost of having an extra device. You can run other functions on top of our WAN optimisation out of the things you would normally run on a server. So in that way it does act and feel a lot like a server, but the things that a server won't do for you is the actual WAN optimisation; dealing with your bandwidth and latency and [N] dedupe and compression pieces. So it gives you the benefit of being able to have that all integrated together and working in a seamless way so you get the optimal amount of performance.
Okay. The other thing I think I'd worry about if I were a potential customer is lock-in.
Basically what you’re saying is, in order to achieve the server consolidation that I want to achieve, I need to buy in to your technology, your proprietary technology, and then do I have an exit strategy if it all goes wrong? Basically what you're saying is Riverbed is a company you can, effectively, bet all your servers on. Why do you think you're good enough to do that?
Well I think it's probably not so much why I think; it's maybe why our 17,000 customers think that we're good enough to bet on. We're a very strong company.
We've been in this business for over 10 years so we're not a start-up. We did over $700 million in revenue last year. So again not like we're an early stage company.
We have many loyal customers. I think it's a reality of IT; when you want the best-inbreed technology, you decide if that's where you want to go. The minute that there's another option, customers could decide to do something different. But I think there's a couple of things that we look at. One is we have a lot of very loyal customers that come back for repurchase with us, that are confident in our ability to sell and support our technologies. I would say, just speaking from our experience with our installedbase customers, that they see us as a very strong and viable solution for them.
Is 17,000 a lot of customers? Doesn't sound like a lot of -- given the number of enterprises there must be out there?
It depends. If you think about the number of independent companies or stores that don't rely on the WAN, then it seems like a small number. If you look at the number of customers that do rely on the WAN, then it's a much bigger number. If I put that in context of what Gartner says is the WAN optimisation market and our market share, we have over 50% share of that. So, based upon the people that are buying this technology, I would say yes, it is.
Okay. I'd like to pick up one point you made in your presentation. You said VDI, virtual desktop infrastructure doesn't save money. How did you arrive at that conclusion?
That was feedback from our customers that they said that they did it because of the control and not the cost savings. So not necessarily that it costs them more money, but the ROI was harder to justify because it didn't have very clear and explicit savings.
One of the things that we strive for is to keep within that payback window of six to nine months. As we designed this technology and looked at the benefit, we wanted to make sure we stayed in that window. The feedback we had from our customers as they looked at the technology was that they were thrilled with that because they didn't realize that was VDI. So it's based upon their direct feedback.
Okay, thanks. Okay, let's see if we have any questions out there. Anyone have any questions for Carolyn?
John Rollason, NetApp
John Rollason from NetApp. Citrix has a WAN optimisation box. It looks almost the same what you do. So could you compare them, if you know them, the Citrix WAN optimisation box which is also --
The Citrix WAN optimisation box, yes. We do see Citrix out in the market. Citrix tends to generally accelerate just to their own technology. So, if somebody is looking for acceleration in a Citrix environment, many people are very happy with that as a solution. What we find is that we accelerate not only Citrix traffic but accelerate everything else. So if people are interested in accelerating their Microsoft applications, the Citrix box would not necessarily give them all of the functionality to be able to do it. So I would make the comparison a very point and specific acceleration versus very broad acceleration that you would get with WAN optimisation.
Any more questions. I have one more. How do I get out of VSI if things go wrong?
What do I do?
If you want to go back from VSI? I think it's pretty -- just like anything, you can turn it off, right? If you turn it off, then you would go back and put the servers back into your branch office locations and you would go back to the operations that you had before. I think one of the reasons people rethink going back to that is, in this environment if you want to restore operations, let's say in a catastrophic failure, and you need to put something in to re-provision a server versus re-provision a WAN optimisation box, it is exponentially more complicated. I think if people looked at that, they would look at it and say, even if things were an equal comparison between having the servers and having this, just that even restoration of operations is exponentially better with this Granite type of a device, that they may actually even rethink the value. But again, it's an architectural choice and some people may decide, for whatever the reasons, to go back. In that case they would just put the servers back in.
Okay. A question here? Yes.
Andreas Lemke - Alcatel Lucent
What would be the typical applications that could benefit most from the VSI? If you take for instance email, email is an inherently network application. So having a cache to continue functionality locally when the WAN breaks down doesn't help you very much. Also you have lots of other applications that need very large data stores that are just too big to be cached locally. So where is the sweet spot? What are the best applications that can most benefit from this VSI technology.
We do find that a lot of the Microsoft applications where people want access to their data for PowerPoint or Microsoft Word or other things like that, SharePoint is a big one especially where, if they didn't have access to that SharePoint information then it would delay or stall work. So SharePoint is a very large driver of what people would use.
Anyone else? Well I think I'm done to be honest. And if no one else has got any questions we could move on. I would like to say thank you very much Carolyn for being a sport and thanks for [indiscernible].