By VSM News Staff published: Saturday, July 24 2004
Surgient is a leading provider of on-demand applications that enable companies to increase the effectiveness of their direct sales, online marketing, technical training, and QA/Test processes. Surgient applications enable the use of distributed enterprise applications by any user, on-demand, from anywhere at anytime by automating and eliminating the complexities and high costs of hardware configuration and software environment provisioning.
This month VSM discusses the key role server virtualization technology plays at Surgient with Erik Josowitz, VP of Marketing, responsible for product and marketing strategies and activities worldwide.
VSM: Erik, could you set the perspective by describing briefly what Surgient does, your role, and how server virtualization fits into the picture?
EJ: I’m the Vice President of Marketing for Surgient. Surgient is an on-demand application software company that is based in Austin, Texas and venture funded. We’ve been around since 1999 and started shipping product in May of 2003. In April of 2003 we acquired the company ProTier. To date, we have 16 customers (including Microsoft, Documentum, Siebel, BEA, and Vignette) and 25 live deployments of our solution. Our customers have delivered more than 200,000 applications on-demand and our data center is managing 1,600 virtual resources across a segment of our customer base today. So we are a large application provider and a large user of virtualization technology as well in the context of the applications we’ve delivered to the marketplace.
Surgient is not a virtualization technology provider. Rather, our four applications sit on top of either VMware or Microsoft Virtual Server and manage specific business processes in the context of software delivery using virtualization technology.
VSM: Tell us more about Surgient’s applications.
EJ: When we look at the world, we look at it primarily in the context of what happens when a company develops software and then has to deliver it to its consumers. That could be a software company, or it could be an IT organization. But if you’re developing software today, the fact is that everybody will tell you that it still takes too long.
The process of writing code is actually pretty efficient today. It’s so effective that you’ve got distributed teams, offshore development, open source development, where the developers never even need to meet each other.
So the headache now in the software development cycle is not writing the code. It’s that, for example, the QA processes take too long. It often takes a minimum of a third of development time, sometimes longer than that, when you have products that have multiple permutations that need to be tested. Then, as another example, handing an enterprise software product off to sales, you’re talking about making the sales organization successful in demonstrating and selling it in the field.
These delivery processes are now the long pole in the software development cycle. The QA process is gated by the resources that are available to it. No company can afford to invest in the resources that would allow them to run all their tests in parallel, because 90% of the time those resources would sit idle in the context of the enterprise.
To take another example, if you’re looking at helping get end users trained, in some ways it is made much more difficult with large software applications. Typically you have to wait until the application is developed and rolled out before you can start training people on it. Best-case scenario is that every student would get a live, hands-on lab, in the context of both classroom and online training. It never happens. Why? Because making the computing resources available cannot be done cost-effectively.
So our overarching perspective is that if you can address those problems without having to force the company to ramp up in computing resources, then you’ve really got something. Because then you can deliver more training more often, you can allow products to be fully tested, sometimes earlier in the development cycle, you can enable more effective sales and marketing of complex applications where end users actually get their hands on the product much more often and thus cause sales to occur much more often as well.
That’s where Surgient comes in. We’ve taken concepts like virtualization and RDP and remote desktop connections and a mid-tier system platform that we call the Virtualization Control Server (VCS) that can effectively orchestrate all of these things to automate today’s complex software delivery processes. We have implemented four process-automation applications that utilize the VCS to deliver complex software applications on-demand: VQMS is for QA and Test, Virtual QA Management System. VDMS (Virtual Demo Management System) is for Sales and Marketing, so sales organizations use it to deliver demonstrations, while marketing organizations use it for online product evaluations. VTMS (Virtual Training Management System) enables companies to deliver hands-on labs for instructor-led or self-paced training – in a classroom or online.
Effectively what we have in our systems is an underlying platform that contains all of these capabilities which are exposed as APIs using open standards such as SOAP and XSLT. We then develop role-based, business process automation applications on top of that, workflow, specific UIs and the like. The same underlying capabilities power all of our applications. They are different from one another more in the specific roles and workflows that a person goes through in order to access the system resources, and the user interfaces are driving that. So scheduling a training class is different than scheduling an online demo. But they both require scheduling.
VSM: Can you comment more about the importance of virtualization in that context?
EJ: It is in some ways one of the foundations that makes this cost-effective. But, at the end of the day, it is really the ability to automate these business processes that is making these customers successful. But the virtualization is in fact the thing that today allows me in my data center, or my customer in their data center, to accomplish this magic in a way that is cost effective. It simply would not be practical to solve these business problems using physical resources, which is why we believe the market opportunity for what we’re offering is so great.
VSM: So it’s an excellent example of how virtualization is being used to deliver specific business value. What you do is enable a pool of resources to be available 24x7, and while it could be done without virtualization, virtualization makes it more efficient, effective, and cost effective.
EJ: Absolutely. And more flexible. Some of the things that we can do with virtual resources, such as taking a software application that is deployed in a virtual environment and cloning it, can’t be done using physical resources without performing manual post-deployment tasks. We have developed patent-pending technology that enables a software environment to be cloned “as is” multiple times without any manual intervention and no network conflicts.
If somebody wants an environment where they have access to an application consisting of a database server, a web server and an application server, we can pull those out of our managed repository and instantiate that as three separate virtual systems virtually networked together.
If somebody then wants to take that web server and turn it into 20 web servers, we just duplicate it. And our system knows that is happening, works out the network issues, so then we can have a 23-machine networked system there for whatever purposes, testing or otherwise.
So the things that we can do because we’re in a virtual environment go well beyond just those cost savings. It also allows us to take specific actions on those resources that would be very difficult (if not impossible) to do utilizing physical resources. You could do it with wiping a system, then re-imaging it, then using SYSPREP or something like that on a physical machine to accomplish what I just described, but it would take hours, sometimes many, many hours, to do. In our case we can do it within minutes, with a couple of mouse clicks, because it’s purely a software operation.
VSM: Can you elaborate on the roles that VMware and Microsoft Virtual Server play at Surgient?
EJ: What we have is VMware or MSVS sitting on top of a pool of physical resources. They can take those physical resources and turn them into n number of virtual resources through the abstract technologies that they provide.
We have an abstraction layer on top of that and it allows us not to have to really worry about whether it’s a Microsoft or VMware virtual machine. That is, the abstraction layer provides a set of vendor-independent services (exposed as APIs) for manipulating virtual resources. Below this layer, our platform uses the VMware or Microsoft APIs that provide roughly the same functionality. Our system then orchestrates the behavior of virtual resources in conjunction with deploying, configuring, and tearing down complex software environments.
One example of this is we can allow someone to take a system that they are working on and snapshot it. So an end user that is working in a QA lab environment may have a failure of some sort while they are running a suite of tests. We have a one-button snapshot that they can click that will basically take the environment as it currently stands, grab the deltas off the Gold Master image that they started with and save those differences back in the managed repositories. Now, when the developer responsible for the component that failed wants to see what happened, he can come into the system and restore the environment. In the meantime, valuable testing resources were not tied up and the QA engineer was able to continue with their tests.
We can do this with multiple systems at once, so there is a scheduling workflow and repository management system that goes along with this that allows us to orchestrate not just single systems but complex clouds of systems.
For the Microsoft Server 2003 TechNet marketing program right now, you can participate in a live lab environment. This allows you to sign up online, and they send you a user name and password, and all this is managed by Surgient, including the outbound email and registration. When you log back into that environment, you get access through your Web browser to one Microsoft Server 2003 system. So you get a desktop, and some instructions in a side panel.
What you really have access to is five machines networked together, configured to show off the latest Active Directory 2003 changes. The evaluation has you go in and change a record, see how the login changes in each of the other four systems, see that whatever server is running on that system also responds to that Active Directory change on that other system. It’s pretty cool.
VSM: It seems like the abstraction layer is the layer that turns the virtual machines into a pool of resources.
EJ: VMware and Microsoft can create individual virtual resources. They are not yet in a position where they can easily create two of them next to one another and then configure them to talk to each other on a virtual network. That’s one of the things we do at that abstraction layer.
Then once we have the ability to do that, there are a whole set of services around what you do with that capability. How do you use it to create the orchestration of one or more machines for an individual just-in-time? How do you reserve and dynamically allocate capacity on this underlying physical infrastructure according to business policies such as the user’s group, or role, or the software application being requested? The mapping of how much physical infrastructure is required - we take care of all of that. We do the resource management.
VSM: Sounds a bit like the old days of “time sharing”.
EJ: It is, in a sense, though what we’re doing is taking the time share context and putting it to work in the context of delivering a software application, let’s say for field sales demos, or to drive online product evaluations. How do you drive classroom-based training, hands-on labs or online virtual labs in the context of a self-paced course? How do you drive the necessary resources to drive through a QA cycle?
VSM: Let’s talk a little more about the business applications. It seems like Surgient is playing a major role in the area of training, demos, that type of knowledge transfer. Would you characterize the market in this regard as being in its infancy?
EJ: Absolutely. And our customers today are for the most part customers that are pegged at the end of the scale in terms of seeing new technology as a competitive advantage. I believe that with the virtualization market today, if people look at it and think how it is going to impact the business world, the first thing they come to is server consolidation. This is where the IT mainstream is right now.
In parallel, and of course these things always play themselves out in parallel, the business units still need to get things to market more quickly, they don’t have computing resources, the budgets are still locked down, IT is still trying to figure out how to fully comply with the CRM system or whatever system it was that they bought last time. They are not available to help do this.
How, in the case of a training organization, are they going to provide hands-on environments for 300 students around the world whenever they need them? Because today’s system of shipping servers around the world is not working. It’s not scalable, the servers get damaged, it requires full-time people to wipe and re-image them. Or in the case of QA the same sort of thing – IT is not always available to help, so how does the business unit solve this problem with fewer people than they had a few years ago? And yet all of a sudden business is ramping up, they are being asked to support more environments, the software being built is more distributed.
VSM: They are asking to go 24x7?
EJ: Exactly. My sense is that while this is in its infancy, the business units are going to wind up driving it, and then two or three years from now IT will wake up again and want to take ownership. All enterprise software markets develop in this same way.
So this is, I think, the next key wave. But when we talk to our customers, the VP of Marketing, the VP of Sales, the first thing they will tell us is that they don’t know how much they are spending on this. So we’ll walk a head of marketing through what they are currently doing. Do you send out trialware CDs to prospective customers? Do you offer online downloads? How many people that get those CDs or download the software actually install it?
Very few, because the system is big, it requires two servers, it takes an hour and a half to install and configure. And by the time they are done with that they don’t have the energy left to actually try the software out. And how much are you spending on that? Say you sent out a million CDs and each package costs $15 fully loaded. So you spent $15,000,000 on a marketing campaign where 19 of every 20 CDs got thrown away or put on the shelf. And of the remaining 5% only 50,000 people successfully make it through the registration process and try the software enough to give you feedback on the process at the end.
For 1/10th the price Surgient can deliver the same capability, but you are only paying for the successful trials, and you don’t have those up front costs. And by the way, people are five times more likely to evaluate software delivered through a web browser than they are to install and configure a CD. And the reason why is obvious. Why would I risk my system, which is in its usual fragile state of stability anyway, installing some new product that I think I might be interested in, only to have to uninstall it later? And this is assuming that my system meets the minimum system requirements to evaluate the software in the first place. There’s simply too much burden being placed on the end-user with trialware CDs and downloads which is why the ROI of such campaigns is so poor.
With Surgient’s solution, I simply open up a web browser and go to a specific URL, I type in a user name and a password and get live access to one or more systems right through the browser, down to the Start bar, and being able to save things on virtual disks, or look at things that are stored on virtual CDs in that network. I can go to other machines that are virtually networked in.
Effectively what we’ve created is the ability to deliver known good software applications to anyone, anywhere on-demand reqardless of the software’s complexity.
VSM: Where could we find that?
EJ: There are three places today that we’re working with Microsoft publicly, just as an example. We’re doing some things for some of their business units around the world, doing some internal things that aren’t yet available publicly. But for over a year now we have been providing an online evaluation of Visual Studio .NET, and that’s been responsible for more than 100,000 trials to date, and a lot of great data about how cost effective this solution has been. That’s primarily just, “Try Visual Studio .NET Now.”
Also, since then the MSDN group wanted to use it for training too, also around Visual Studio .NET. So they deployed our training application, and now there is training for use with Visual Studio .NET that effectively blends multimedia training content and lab exercises with a live, hands-on environment.
The third thing is the Server 2003 environment, which is made available through the TechNet group at Microsoft. That launched just about a month ago, and I believe they delivered 1,200 trials of that within the first 24 hours. So it’s still early for us to have full knowledge of how it is going to do for them, but they’re just thrilled.
What they found is that for something like Server 2003, which is big enough and complicated enough that almost no one ever installs those CDs that they get, customers and partners finally have a way to actually try a new version of an operating system, evaluate specific features, look for specific changes, without ever having to deploy it on their hardware. It’s just a big win for them.
VSM: If you go to that evaluation, would you have any idea that it involves Surgient?
EJ: Yes, there is a little “Powered by Surgient” link at the bottom there.
VSM: So the person using that has kind of their own little lab? Able to do what they want to do, as opposed to before, to follow a rigid pattern or outline of a CD-based training.
EJ: Exactly.
VSM: So you provide a lot of flexibility, let people learn at their own pace, when they want to. That could have very profound implications for lots of types of training. Are you doing anything in the educational market, with universities?
EJ: We haven’t really started looking at universities, and it has more to do with, as we discussed earlier, the fact that the market for this is early, and universities tend to follow behind a little bit. What we have been focused on in the short term is companies that have to deliver technical competency. So in delivering technical competency there are a couple of ways to go about it.
One is through simulations. But simulations are not going to solve the problem for training people on Install and Config, it’s not going to solve the problem for training people on application development, it’s not going to solve the problem in training people on break and fix of complex systems. So what we have been focused on in the short term are those people, like for example Wily Technology, whose product exists in a technical world and are used by technical users.
People need to know how to instrument an application that they are running in the context of, say, BEA’s WebLogic application server, with the Wily agents. So they can use Wily’s Introscope system management tools then to come back and manage and monitor that. They have to deliver detailed technical knowledge to an audience that is global. Wily is a relatively small company, certainly when compared to companies like Microsoft or Oracle. For them, this meant that they didn’t have to go to the step of building training centers all around the world. Now what they do is they have instructors that exist around the world, all of whom can pull from a central pool of on demand training systems. This allows them to scale like a big company without having to resource like a big company.
VSM: We talked about how this is really in its infancy. Can you share with us where you might see this technology leading us in the future?
EJ: We have a couple of interesting customer examples that start to point the way. The first one is Symantec. Symantec needs to be able to ensure that the latest version of their virus-checking product is effectively getting rid of viruses. But they don’t necessarily want to do that by letting viruses loose into the wild. They used to build isolated networks to do this.
Now they are using our product to build virtual networks where they can let viruses loose and actually watch how they proliferate in the environment. This allows them to have more detailed knowledge of how those viruses act earlier in the development cycle. Then they go back and inoculate those same systems with the latest version of their virus-check product, and make sure that the virus is actually eradicated. So it’s a virtual lab, but in a slightly different way.
The next version of that really starts to point the way to the future. And that’s what we are doing with Starwood Hotels. Starwood obviously builds and manages hotels, and their IT group has a property management system that they have to configure and deploy to their hotel properties.
The problem that their QA organization was having, though, was that they were under-funded – they didn’t have enough resources to effectively test – and that they had to test on the particular computing resources that were going to be deployed onsite at a new property. But they either had to get those in-house, or they had to wait until the property was actually built and those resources were put live.
What this system now allows their QA organization to do is actually guarantee new versions of software much earlier in the process of deploying a new property. It allows the company as a whole to be a lot more nimble.
Another one that is similar is National Instruments. They are a more traditional technology company, like many of our other customers. Dell is doing some interesting things with our products in terms of provisioning new systems for customers as well.
One of the core technologies that we have developed is something that allows us to take a software environment from physical to virtual to physical. And typically that’s tough, as you well know. And I know you’ve written about it a great deal, because going from physical to virtual isn’t in itself so bad. But going from virtual back to physical you have the problem of device drivers and everything else.
What we are able to do is, through patent-pending technology we call ICE, from a virtual image we can pull out all of the hardware dependent capabilities and replace them as we are going back to the new physical system with the correct ones. So we have a catalogue of drivers and whatnot that we work with, and we have tested this on a number of systems and it works fairly well. It’s not as quick a process as we would like right now, but for customers that absolutely need to model future physical environments, test against them, or run applications against them, and then want to roll those out to actual physical resources without having to go and rebuild and reconfigure, which is the typical thing, we can actually do that today.
VSM: If somebody needed that V2P, where is that found, which product?
EJ: We actually do that within the core of the system so all of the applications can take advantage of it. In sales what we see sometimes is an SE organization that wants to deploy demos but ultimately wants to get them deployed onsite on a physical machine because they don’t want the demonstration to have to require Internet access when it is being used. In a QA environment it makes perfect sense of course.
In order to accomplish what we’ve done, we have had to build a number of capabilities on top of the core virtualization engines, some of which we expect over time Microsoft and VMware to provide for us. Again, VMware has recently added VMotion and they are on a path towards ultimately having some sort of V2P2V ability. And that’s fine with us. We’ve had to do it ourselves for our own specific purposes in the meantime.
VSM: Could you share with readers more about your company and its use of virtualization technology?
EJ: As you might imagine, because we host a good number of our customers’ applications, we are probably one of the major data center users of virtualization technology today. And so just to give some scale to that, the way that we look at what is running in our data center is first the number of servers; second we look at how many customer applications are being delivered through those servers; and then how many virtual resources are required to deliver those customer applications.
Our data center currently has 178 servers that are managed using our software. Those are delivering 800 individual applications within that managed pool. And at any time there are over 1,600 virtual resources. So there are 1,600 virtual servers configured for 800+ applications living within this pool of servers at any one time. And we have delivered well over 200,000 end user experiences in this manner.
VSM: As an advanced user of virtualization technologies, and using both Microsoft and VMware, how do you compare the two?
EJ: To be honest, we find them more or less equivalent today. We have been working with the Microsoft product since it was Connectix, and we worked with that team since it transitioned into Microsoft. We find it to be very stable and very usable in its current beta form. It has been pushed off a little in terms of a marketing launch today, but we are running it for Microsoft right now in delivering their applications, and have had a lot of success doing that. We obviously use a lot of ESX, and a lot of VMware solutions, and we are very happy with how that performs for us.
VSM: Do you have a ballpark number of how many GSX/ESX servers you have vs. how many Microsoft servers?
EJ: I can’t really quantify it in terms of number of servers, but because Microsoft Virtual Server is still in beta I believe the only customer we are actually running that for today is Microsoft. And we run a lot in our data center for Microsoft. They are one of our larger customers, as you might imagine. So depending on how you look at it in terms of racks, or in terms of virtual resources, the number might be a different percentage compared to VMware. But the majority of our customers today are using VMware.
VSM: Erik, who are Surgient’s competitors?
EJ: In terms of competition, primarily today it’s build vs. buy. So we see a lot of IT organizations trying to build some of these capabilities using virtualization and then building some custom tools on top of it. We see people trying to do similar things using Altiris, or even Symantec Ghost, to accomplish similar results. Usually what we find is that if we give a customer an idea and they go off and try and do that, usually they come back to us and they say, “We thought we could do it, but this is harder than it looks.”
There are really two primary companies that do similar things to what we do, and they are Toolwire and Hatsize. They are almost wholly focused on the training space, and they deliver virtual labs in the context of training applications. They can only do single server systems for the most part now, so they can’t really do enterprise applications. So they tend to be tools training rather than systems training. And there are a number of other ways in which we are different from them. But there is starting to be recognition at the business level that providing hands-on access to live software applications is something that can’t be lost in the context of all of the various cuts in resource availability, and that this is a way of having your cake and eating it too.
At Surgient we are actually starting with the applications and delivering the infrastructure underneath it. I think what’s going to happen is that you are going to have the virtualization vendors, then the virtualization management vendors, and there are going to be people that build up layers and layers of infrastructure trying to, over time, get to real business process applications.
One of the things I personally like about the Surgient story is that we’re there already. And our customers really recognize it, and they are saving money today, and they are able to quantify it, again, partly because what we are delivering to many of them is a subscription to these capabilities. So they are able to quantify it very quickly and see a very direct ROI on this. So in a sales or marketing situation it becomes a no-brainer for them. Of course they’re going to keep doing that because the alternative is to stop doing that thing altogether, and that’s the thing that is actually driving the business right now. So it’s a fun place to be and an exciting place in the market.
