LEXINGTON, MASS. (05/23/2000) - If you're still not buying the hype about the emerging application service provider (ASP) market, then take a look at how the concept is helping this small software developer reach a larger customer base.
The tale shows how the ASP concept is leading vendors in new directions and opening up opportunities for corporate IT folks.
Exa Corp. in Lexington, Massachusetts, sells a simulation tool that companies such as BMW, Audi and Ford use to study how air flows over and under cars. The company's PowerFLOW software is faster (models can be developed directly from CAD drawings), more accurate (traditional methods use partial differential equations that sometimes return meaningless results) and easier to use (you don't have to be a math wizard) than traditional computational fluid dynamic tools.
The product is a godsend for automakers - Exa's primary vertical market - because it lets them test things such as how much wind noise sideview mirrors will make, without having to build prototypes for use in wind tunnels.
Prototyping and tunnel testing cost money and delay the delivery of a new product. That's why companies such as BMW want to get away from prototyping altogether within five years, according to Exa co-founder, President and CEO Stephen Remondi.
But the downside of simulation is you need lots of horsepower to run Exa's software. Ford and BMW use 100-processor Silicon Graphics (SGI) machines for PowerFLOW.
That kind of hardware investment doesn't intimidate car makers because they are familiar with large-scale processing and can justify the expense, but it is beyond the pale for smaller players. Few automotive parts suppliers, for example, can afford the investment, even though they might benefit from it.
Take manufacturers of windshield wipers. "Something as simple as wiper performance can influence the car buying decision," Remondi says. Design them wrong, and wind will create too much lift and result in poor visibility. If a wiper manufacturer can demonstrate how its product will perform on specific cars, it might gain a competitive edge.
The ASP concept
That's where the ASP concept comes in.
To reach product shops that otherwise could not afford its tools, Exa decided to host PowerFLOW on the Web. After looking at Exodus, PSINet and other options, Exa picked GTE Internetworking (now Genuity) as its Web-hosting partner because of its proximity, capabilities and security, Remondi says. Last November, Exa installed a 32-processor SGI cluster in GTE Internetworking's Cambridge, Mass., data center.
Today, Exa's e-CFD service lets customers upload simulation jobs and pay $7.50 per hour to use the resources. There are no sign-up fees or recurring fees.
Customers simply pay for what they use.
But they don't always get the results back over the Web. Customer input of 10M bytes of data can result in a multigigabyte simulation. Instead of getting that shipped back over the 'Net, customers often opt to get the bulk of the simulation captured on tape and mailed.
"In truck design, you may want to study drag and lift," Remondi says. "Those are simple numbers we can return over the 'Net. But if you want to know what's causing the drag and lift, you need to see the simulation. That's a huge amount of data, and customers will often have that FedExed."
This summer, the company will look at adding support for remote visualization, a technique that would enable Exa to ship images out to customers instead of the 100M bytes of data required to generate the image.
Remondi says customers are responding favorably to the ASP concept. The business already accounts for 12 percent of Exa's revenue and, given current run rates, will represent 50 percent of revenue by next year.
Exa already has plans to upgrade its current 32-processor machine to 64 processors in the second quarter, and 128 processors by the fourth. That will drive down hourly costs even further, Remondi says.
While it is still cheaper for the big automakers to install the hardware and host PowerFLOW themselves, the ASP model opens the door to a world of vendors that until now have sent wares into production untested.
Remondi estimates that only one in 10 devices that could benefit from fluid mechanic study actually gets examined. That can be costly. For example, designing a computer without actually studying how air will flow through it may result in unnecessary heat buildup.
"You could be two years into the life of a product and suddenly see failure rates skyrocket," he says.
Money's an issue
Reducing the cost of entry using the ASP model is "radically changing how customers perceive fluid mechanics," Remondi says.
Oddly enough, cost is working against Remondi as he looks to other ASPs to host his company's business applications. "It costs three times as much as doing it yourself," he says of the general business ASPs he's familiar with. The reason?
Siebel and other software developers won't let the ASPs share licenses across users.
"I may use Siebel one day per week," Remondi says. "The ASP should be able to buy one license and get four more customers like me, then charge each of us for half a license. We would save money, and the ASP would make money."
The industry has to work out some of these arrangements before it can really fly, he says.
It is clear, however, that the ASP model is positively changing the dynamics for vendors and users alike. Consider this final example: Now that e-CFD is up and running, Exa is looking at how it can add additional value. Remondi says that may entail integrating complementary software from other vendors, such as tools that would enable automakers to simulate car crashes.
With the ASP way of doing business the vendor wins, the customer wins and new capabilities are born. No wonder there is so much hype surrounding the concept.