Software Licensing Gridlock
Software licensing models may need to change to support the grid computing movement
May 25, 2004
![Network Computing logo Network Computing logo](https://eu-images.contentstack.com/v3/assets/bltde8121fc52c5c8f3/blt3f3d0318f746b1c2/65a530e4187606040a1d8b8c/placeholder.png?width=1280&auto=webp&quality=95&format=jpg&disable=upscale)
PHILADELPHIA -- Gt'04 -- Software licensing promises to be one of the major challenges for IT managers looking to deploy complex grid infrastructures, according to attendees at this years Gt’04 conference in Philadelphia.
Jeff Mathers, senior consultant in Johnson & Johnson's research innovation division believes that the move to grid computing requires a change in how many application vendors license their products.
He says, “In many cases, they have to move from a model based on a large CPU device, to a more distributed model that could be based on a [specific] site or even regions.”
Much of grid computing’s appeal centers on the fact that it enables companies to harness server, mainframe, and even workstation resources across a wide area. Typically, these deployments employ complex virtualization technologies, and applications need to be shared across a broad range of computing platforms.
Johnson & Johnson has already implemented a grid infrastructure based primarily around Hewlett-Packard Co. (NYSE: HPQ) and Intel Corp. (Nasdaq: INTC) machines and Sun Microsystems Inc. (Nasdaq: SUNW) Solaris servers to support its research. To date, the company’s grid is capable of scaling up to 600 nodes, although this infrastructure will become increasingly complex over the next 18 months. Mathers estimates that the grid could even number several thousand nodes by the end of 2005.The challenge for Johnson & Johnson, like many other pharmaceutical firms, is how to deal with a large number of small vendors providing specialized applications.
Johnson & Johnson currently deals with around 15 such companies to support its grid computing strategy. But at least one of these has moved from per-user licensing to a model focused more on specific company locations, according to Mathers.
But it’s not only the smaller application vendors that pose a problem when it comes to developing new software licensing models for grid computing. Mathers says “I think that there is even less leeway with the larger vendors.” Sadly, he refused to name any of these firms.
Among Gt’O4 attendees, Mathers is not alone in expressing this headache. Another IT manager, who asked not to be named for fear of damaging relations with his suppliers, says, “I think that everyone is having software licensing problems.”
Most IT managers that Next-gen Data Center Forum spoke to cited the sheer complexity of grid computing as the catalyst for their licensing hassles. In a typical manufacturing grid, for example, the computing architecture could encompass key suppliers’ IT operations.F. Brett Berlin, adjunct professor at George Mason University’s school of computational sciences, sees the design space as a licensing nightmare. “Large simulation modeling applications will have to have new kinds of licenses that can go beyond the geographic site,” he says.
This will place an additional strain on the application vendor, according to Berlin. “If a company has to have an application running on its partners’ computers, for example, the license will have to go across geography as well.”