Customers won’t need to completely rewrite their software for it to work with the utility computing model, according to some vendors. But one developer says the grid environment is so different from the desktop that developers will still have to rethink how they write applications, while porting existing applications will take a significant amount of work.
Benny Souder, vice-president, distributed database development for Oracle Corp.’s server technologies division, said for the most part, enterprises will be able to take their existing software investments and move them into a grid environment without too much trouble.
“It’s not a rip-and-replace kind of thing,” Souder said. “You can start with a small number of components, move a couple of things, gain more experience and confidence, and then move over more components.”
He cited a study from Framingham, Mass. analyst firm IDC Corp. which found that most IT departments are already involved in some aspect of grid computing, mostly around dynamically provisioning storage or servers.
He said there are various toolkits available to help enterprises move their applications over to a utility environment, and “all that is good if you want to use that stuff, but it’s not necessary to rewrite (the applications).”
However, Tony Enerson, chief technology officer with software development firm Catalyst Realtime Corp. in Calgary, was more hesitant about the ease with which apps can be adapted to a grid environment.
Enerson said application development for the grid is inherently different from regular development. Applications meant for the grid must be “(broken) apart…into medium-sized pieces that have to be portable so they can execute anywhere,” whereas non-grid-enabled apps are designed to respond to one user’s keystrokes and run on one CPU.
Porting applications from a standard data processing environment to the grid is a “really big deal,” Enerson said. “It is a redesign of the application.” To accomplish the task, developers will need a refactoring tool, which “changes the way the software is inside without changing the way the software works on the outside.”
For example, a refactoring tool may enable a developer to break an application into two pieces while still allowing the user to just see one. “Those tools will pave the way to take server applications and make them grid-enabled,” he said.
Enerson said standardization is also an issue for applications that run on the grid. “Right now, applications for the grid are grid-dependent, meaning they will work with this grid technology but not with this other grid technology. People who are writing applications must choose their grid very wisely because they’re going to sign on with that grid for the long haul.”
Markham, Ont.-based Platform Computing Inc. provides platform-agnostic enterprise grid software and services that allow organizations to plan, build, run and manage grids by optimizing IT resources.
According to Platform’s product manager Marnie Biles, Platform’s solutions can simplify the integration of an enterprise’s existing applications with a grid environment. “It really varies depending on the type of application and the degree of its complexity,” said Biles.
“But it typically does not take very long. Once you have one application done, they are quite similar. In some products there is very little or no work at all.”
Depending on the nature of the customer’s existing software, they can choose one of Platform’s grid and workload management offerings. If the customer runs compute- and data-intensive batch workload processing applications, it can use Platform LSF.
For financial and other customers using primarily business-oriented applications, Platform Symphony is the best choice, she said.