Fellow ZDNet blogger David Berlind describes events at the recent GlobusWorld gird confab in Boston.
Grid represents the next generation of clustering -- in which workloads can be spread across multiple processors -- and much more. Unlike clustering, you're not bound by a single platform. Processing power can be borrowed from any type of machine -- from desktop computer to mainframe -- to accomplish a job. David quotes grid researcher Ian Foster, who states that "computingconcepts such as Web services, utility computing, virtualization, datacenter automation, and the adaptive enterprise are all solving the sameproblem and should be solving it in the same way."
Indeed, it would seem natural, if not inevitable, for Web services/SOA to fuse with grid into a single approach for dynamically and automatically managing computing workloads. Both Web services and grid services focus on supporting applications or processes running across multiple loosely coupled systems. The success of both approaches rests on availability, performance, and platform independence.
However, grid computing and Web services exist in two separate worlds, each with its own proponents and agendas. While Web services has been highly commercialized and can be executed within the smallest systems, grids are seen as huge, sophisticated networks mainly for high-end scientific or engineering applications.
Everyone wants Web services, which can leverage current systems and cut down development time through code reuse. Everyone sees the value and potential savings in grid, which distributes processing across grids of already existing systems, versus buying new systems. At this point, it's unclear exactly how many organizations will want or require grid services standards within their infrastructure.
IBM Corp., along with the Globus Alliance, HP, Akamai, Sonic Software and again, Tibco, have put forth a set of Web services specifications, WS-Resource Framework (WSRF), which is built on the Open Grid Services Architecture supported by Globus Alliance. WSRF employs Web services to identify and leverage underutilized capacity in both physical and logical resources from across the network.