GT: Welcome to ebizQ's "First Look" podcast. I'm your host, Gian Trotta. BEA co-founder Bill Coleman has a new company that's in the forefront of the "Cheap Revolution." Cassatt Software is part of a surging trend of small, start-up companies are using Open Source and Web-based software that runs on less expensive hardware to undercut the big vendors. Cassatt's recently upgraded Collage program controls a pool of servers and other network resources to meet peak demands. Companies like Pfizer have already become customers and are using Collage to cut their hardware and IT costs. Here to discuss the trend is Ken Oestreich, Cassatt's Director of Product Management. Welcome, Ken and thanks for joining us.

KO: Thank for having me! I appreciate it.

GT: Ken, a quickie to start--what's the origin of the name Cassatt?

KO: Funny, we get that question a lot, it’s actually named after Mary Cassatt, she was, I believe a 19th century American impressionist, and a revolutionary in her own right, so ergo we adopted the name.

GT: Okay. Well, we'll go off to Wikipedia to see if you also wrote the entry.

KO: Hopefully, I'm not lying!

GT: (laughs) OK. Second question, Ken. Ken, what real world benefits are customers like Pfizer gaining from Collage?

KO: Okay, yeah--a lot of people using this sort of solution. Briefly, it’s a utility style solution, so two real big areas of benefit have to do with capital efficiency, you know--how efficient your hardware is being used because what we do is help pool your resources, any kind of resources whether they’re virtual or physical so you get better utilization of them, and then on an operational side of things, we automate a lot of the day-to-day firefighting and provisioning and scaling, so that your operational resources are at times more efficient.

So ultimately you kind of get out of it and our customers have told us this, a good deal of agility out of this, for example speed to deployment. Not so much a need to do traditional capacity planning anymore because the system is taking care of it for you in sort of real time consolidation. And then the last thing, kind of an interesting thing, is power reduction. The machines that are not being used as part of the pool are actually being turned off. So, again, a lot of those operational benefits and clearly the capital benefits.

GT: Right. Are there advantages and redundancy also?

KO: Absolutely! Redundancy not in the traditional sense, redundancy in the form of spare pool equipment so that when you need capacity or if there is a failure, bare metal machines can be turned on, and reallocated in a matter of seconds or minutes.

GT: Okay, great! I wanted to go a little deeper into the actual process. How does Collage's virtualization capabilities manage to get previously siloed servers to work together at peak load times?

KO: Ok so a little bit of background, we’re not a virtualization provider, per se, but we do manage existing things like, hypervisors, and we do virtualize to the degree that we can decouple hardware and software and network so what it gives you in terms of the traditional siloed message of computing is the ability as I said, to pool the physical and virtual resources, then when you’ve done that, what our system does, is it attaches an SOA service-level agreement, an arbitrary one that you define to each application. And then after defining that you deploy the application into the pool and it in effect manages itself. All right?

So, take for example, during peak times, if an application is out of service-level agreement, Collage, which is the project we’re taking about allocates additional free bare metal resources to that service, and actually has enough intelligence, for example, that if you actually run out of resources it can actually reallocate resources from a lower priority application to your high priority application, so this concept of siloed systems completely disappears.

GT: That's very interesting. Ken, what are some other advantages of this kind of virtualization in regards to performance, software selection, and business agility?

KO: Oh boy, performance software and agility. From a performance perspective, there is actually no performance hit taken to give the applications. We don’t employ any software layers, we don’t employ any agents on existing servers. We set off to the side as kind of a universal remote control if you will, so there’s actually zero performance hit to existing applications.

From a software perspective side, we’ve chosen a path to be, completely OS and platform and software agnostic. So you don’t change the software, you don’t modify your software, you use the stacks and the tuned software that you're accustomed to using. You shouldn’t need to change anything. But the real neat thing is this last thing you brought up, business agility, which I look at a lot of analysts and they say this is actually the number one concern, probably next to cost. Clearly this pooled, automated approach, gives IT management a faster time to market, a faster time to deploy. Because you're in effect, dropping an application into a pool of resources. So you don't do traditional capacity planning, though you'd have to. You don't need to do traditional consolidation planning, the way that you're accustomed to. Because you're sharing resources.

And then the real interesting thing here is, the way it integrates with things like asset management, compliance management or business process management, BPM. Because you can actually use, let's say, an existing BPM system to help govern the service-level agreements that are controlled by our system. So, take for example--a BPM system that says, we know the certain set of applications that produce a particular business process for you. We notice that one of those is out of SLA. The BPM system can tell our system, Cassatt Collage, perhaps to predictively access and apply new resources to those existing processes in advance, perhaps, of their getting busy or overloaded.

So there's actually, you'll see a nice synergy, between things like BPM systems and utility computing and automation systems in the future.

GT: And that's something I hope you'll keep us posted on. Another possible facet--how about SOA governance and testing? Is there any application there?

KO: Governance and testing, yes--to a degree. Governance in the form of the following. Not so much from a software modeling perspective. Remember, Cassatt Collage operates at the hardware infrastructure level. Most people think of SOA at a software level. At a software architect level.

GT: Yes.

KO: But there's this concept of service-oriented infrastructure. So that what we hope to achieve, we've achieved it, we hope customers achieve, is an infrastructure that adapts to the SOA component needs. So it's impossible always to predict how many SOA components you've deployed and what their load levels are going to be and what their demands are going to be. But each SOA component does have an SLA attached to it. And you might have hundreds or thousands of these in an environment. So from a governance perspective, we can automatically adapt the hardware infrastructure to continually support the sometimes unpredictable demands of SOA components and composite applications. So there's a beautiful, beautiful, complementarity between the service level automation that we do and SOA.

GT: Wow. Ken, this is fascinating, and I have to give you many thanks for joining us. I hope we can follow up on both those attributes of BPM and SOA in the future.

KO: Well, it's actually my pleasure. It's actually a cool time for all of these technologies.

GT: This is great. I know that, you know, we obviously want to lead with your core business and core application, but the whole concept of the SOI, the service-oriented infrastructure, it would be great if you do move in that direction, would you please--through Steve or yourself--let me know?

KO: Yeah. We'll let you know. And I'll tell you, we've had conversations with people like AmberPoint, for example. If you're ever in touch with them. Where, you know, picture things like their Enterprise Service Bus. Which is continuously monitoring all of the SOA components. And it knows what SLAs are for each of these components. Right? So there's--again, as I said--there's a complementarity between their knowledge of the components and their SLAs, and our ability to affect change in the hardware. So I suspect you'll see these things growing closer together in the future.

GT: It certainly is. Everything is interconnected.

KO: Yeah, it's a really rich area.

GT: Yeah, it really is. I mean, each one in its own is being heralded as the next best thing. Each one is considered revolution and--

KO: (laughs) I know! And you wonder how much of it is real sometimes, right?

GT: Right.

KO: But I think the proof will be in the pudding, when you start seeing how many customers start using the stuff.

GT: Right. And I think that the difference is right now, it hasn't become a commodity and price is not as much as object as performance.

KO: Yeah, I'll tell you. So that's really true. I mean, think about this for a minute: the capital costs in the average data center? Are somewhere into 30 and 40 percent of their annual costs. Operational costs are like 50 to 60 plus percent. So it's not the cost of stuff. You know, it's not the cost of things anymore. You know? It's either the value they bring or the operational costs they cause, that people are looking at. You know, or in our case, right--it's gonna be the operational costs that they save.

GT: Right!

KO: I think you're going to see a huge amount of, you know, of savings on the operational side. People don't care what the thing costs--it's what the thing does for you.

GT: Right. I think that a lot of technology investments. You know, people weren't able to quickly demonstrate ROI and then derive a clear benefit. I know in the BPM space, a lot of people are cherry and a come-from-Missouri attitude? You know? But I think that your value proposition is very clear. You don't have to keep a big server for them. Like maybe, two counties away? Where land and electricity cost a little less but then when things go wrong, you have to constantly shift your workplace, your work force, back-and-forth, back-and-forth.

KO: Right, right. I mean, if--it's certainly an overused analogy, but I mean, electricity actually really works for this, right? Cause you pool generation facilities and no matter how big the load you have hanging off your wire is, the appropriate amount of electricity flows to it, right? And if your 100 watt bulb turns into a 10,000 watt lighting system for a stadium--the electricity sort of flows to meet the need. Right? And somewhere upstream, you know, the Hoover dam opens up an extra generator. The analogy is the same. You have an application. It goes from, you know, web site goes from a thousand hits to a million hits. You just open up another generator. You turn on another machine. Automatically, to accommodate that. And compute power is definitely, definitely going in this direction.

GT: And it's a good way to go, obviously. I can remember, well, I was at Time Inc. when the first Sports Illustrated swimsuit issue went online?

KO: (laughs)

GT: (laughs) Exactly what you described!

KO: You know, it's funny, the example I give is the couple of Super Bowls ago, the Victoria's Secret catalog roll-out during halftime? And it bought their servers to their knees. It really happens that way.

GT: Yeah, and it is the kind of thing, wow! We should have worked that in. Anyway, Ken. I can't thank you enough!

KO: Good deal!

GT: Okay, great! And--

KO:

GT: Oh, great!

KO: I have a commute and it's good to listen to these things.

GT: Well, thank you. Thank you. That's what I tell myself, I listen to, you know, other podcasts on the way. It's a good time to drive…

KO: Exactly.

GT: Okay, thank you so much, Ken.

KO: All right, sir! Have a good afternoon.

GT: You too.

KO: Bye.

GT: Listeners wanting to learn more about Cassatt can go to www.cassatt.com. For more cutting-edge virtual conferences, podcasts, Webinars, white papers and more, the address is www.ebizq.net.