Translattice is one of those rare companies that could turn the computing world upside down. Its plan is to change computers so that they process applications in the places that are closest to where they are being used. It sounds simple and logical, but the biggest enterprise applications are still centralized in big data centers.

Big enterprise technology companies talk about how they are embracing the cloud. But Translattice’s founders, Mike Lyle (pictured left) and Frank Huerta (right) believe they aren’t going far enough. So in the past three years, they created a distributed application platform. Their goal is to knock 60 percent to 70 percent off the cost of running a global application such as a corporate email system.

Translattice’s software runs on standard Intel data center computers known as servers, with eight processors, a bunch of hard disks, and a lot of memory. Translattice sells these servers to enterprises, which sprinkle them around their various geographic locations.

Those servers are like the building blocks of a utility (utility computing has long been a Holy Grail of the tech industry, where you can turn on or turn off computing power with the flip of a switch).

With Translattice, you get that. You add more computers, you get more resources. You take one away, you have fewer resources. But nothing crashes or goes down. The computing load is not centralized in one place; it’s distributed across the network.

“The big story about Translattice is that they appear to understand the infrastructure implications of the shift to mobility remarkably well,” said Elisabeth Rainge, an analyst at market researcher International Data Corp. “Very very few suppliers understand this. Even fewer have a solution that is in keeping with the massive architectural shifts.”

By contrast, corporations today are running a lot of centralized software (see above illustration), even though the users who tap into it are based in other countries, are in other geographic locations, or are on the run. Even as they launch software-as-a-service applications — which tap the power of the web to allow users to log in from anywhere — they still suffer from outages and slowdowns that result from too much complexity.

“The monolithic approach can’t handle the demands of today’s enterprise,” said Lyle. “The current infrastructure is too centralized, even as companies operate globally.”

It takes special software to do this, and that is what Lyle, chief technology officer, and his engineering team have spent the last three years building. The software will take something like a giant Oracle database and subdivide it so that it can be distributed across a bunch of computers.

“Our magic is to tie it all together,” Huerta said. “This has been a problem for 30 years.”

The magic of this kind of platform is that it distributes the database around the company. Normally, companies have to maintain a giant data center that is backed up to another data center. But with Translattice, that backup data center is no longer necessary. And that’s where some huge cost savings start to accumulate.

If you want the network to cost less, you can change the policy (rules such as minimum performance levels) for the infrastructure and pull out servers. But if you want to have more reliability, you can set the minimum threshold for hardware. Since more computing happens close to where it is needed, networking and storage costs go down. If one node fails, you lose a portion of your ability to process an application, but the application itself doesn’t go down because it is being processed in many other nodes (see second illustration).

The company has rolled out this big idea slowly. It announced itself in August and described its platform for the first time in November. It is testing the platform with a few companies now in a beta test and it plans to launch the full platform in the second quarter of 2011. So far, the reception has been good.

Some analysts have sung its praises, although there isn’t a huge body of analysts who have weighed in. Rainge at IDC says that the new mobility of workers will require a lot of new computing and networking capacity at big corporations. Translattice’s solution could be a key starting point for what will be a big information technology makeover, Rainge said.

“I will wait to see product, but the architecture is good,” said Anne MacFarland, senior contributing analyst at the Clipper Group, a technology acquisition consultancy outside of Boston. “They picked a good place to attack, focusing on relational databases. It’s a huge market opportunity.”

Lyle says that big banks or corporations with dispersed sales organizations will be able to tap Translattice and reap big savings. Running a customer relationship management app or a corporate email system across all regions of the world won’t be difficult to do.

The team is interesting. Huerta and Lyle worked together a decade ago in a successful startup, Recourse Technologies, a security technology firm that was bought by Symantec. Recourse made “honey pots,” which were traps where enterprises could corral hackers and watch what they were trying to do as they broke into a company’s computer network. I wrote about them in the Wall Street Journal back in 1999. Symantec bought them for $153 million in 2002.

Huerta went on to become chief executive of medical devices firm Cartilix, while Lyle stayed for a time as chief architect at Symantec and then became an entrepreneur-in-residence at DCM before starting Translattice in 2007. Huerta and Lyle are joined by a number of other Recourse employees, including Robert Geiger, who is also a co-founder and is serving as vice president of engineering. They have 20 employees and have raised $9.5 million from DCM. The company expects to raise a new round early next year.

There are huge obstacles to execution. Big corporations tend not to trust little startups with great ideas. So the testing process will be drawn out. And the biggest server vendors — Hewlett-Packard, Oracle, IBM, Dell and others — all want to own the market for this kind of hardware and software.

“It takes a lot of missionary work,” Huerta said. “We have to engage with customers early.”

Fortunately, Translattice doesn’t have to get all of the $50 billion enterprise computing market to win. It can do just fine with just a slice of it.