With the first beta for Microsoft Corp.s upcoming Windows Server 2003 Compute Cluster Solution barely out the door, the development team is already thinking seriously about features and functionality designed to make the next version of the product even easier to use and manage than its Linux competitor.
The feature-complete beta for Compute Cluster Solution, a 64-bit operating system for industry-standard x64 processors, was released late last month. The product is designed to address the HPC (high-performance computing) needs of Microsofts enterprise and corporate customers, particularly at the departmental and workgroup levels.
But Microsoft officials know they have some catching up to do if they intend to lure customers away from Linux, which dominates the HPC world.
"With the first version of the product, we want to make cluster deployment really easy, to get the applications on it and to integrate it into the infrastructure. For [the second version], we are already thinking about clusters of clusters," Kyril Faenov, Microsofts director of HPC, in Redmond, Wash., said. "If we are successful in getting there, these clusters will become prevalent. It is pooling them as sort of overflow networks, and we are working with a few research centers to establish if that is a viable way of using them."
The development team will continue to invest in making clusters easier to manage, particularly as their use moves away from centralized resources and down to the workgroup level. Microsoft also plans to work on improving the user experience through better power management and use of the IPMI (Intelligent Platform Management Interface) or Web services management to reboot clusters without having to power down, Faenov said.
"[The second version] will look at how we can control the hardware using the standard ways of rebooting. We will also look at how we can make the integration of clusters into the applications, including workstation applications, even easier. We are also looking at performance monitoring tools for the cluster space as a whole, which is a big area for development," Faenov said.
Some users, such as David Robert, a systems manager for a global consulting and engineering company in Cambridge, Mass., are confident Microsoft will be successful in making a product that is easier to use than Linux.
"We are in the process of setting up a three-way Oracle [Corp.] cluster on Linux that uses a shared file system, and it has been a bear, what with the lack of documentation and problems with our backup program not being compatible. Im sure Microsoft will do a better job in this arena than Linux," Robert said.
Microsoft is also interested in data storage. As users begin to run jobs across wide areas, scheduling such tasks too far away from related data can be time-consuming and expensive.
While Microsoft executives, including CEO Steve Ballmer, have admitted that Linux is ahead in the area of HPC, they are looking at how to offer customers an even better solution.
But not everyone is convinced the company will meet this goal.
"I do wonder how Microsoft thinks they can offer a product that competes with one that includes security enhancements from the National Security Agency," said Charles Kramer, chief technology officer at Social & Scientific Systems Inc., in Silver Spring, Md. "I can easily agree with the degree of difficulty to set up and manage existing clusters, but I cant imagine a secure system that wasnt difficult due to the ongoing war with hackers and malware."
In a nod to the pervasiveness and success of open-source software in the HPC space, Faenov said that Microsoft is including the open-source MPI (Message Passing Interface) technology in its Compute Cluster Solution. MPI is a library specification for message passing.
Microsoft is working with Argonne National Laboratory, a U.S. Department of Energy laboratory operated by the University of Chicago, and has taken its MPICH2 implementation, which most ISVs have tested code against, and optimized it for performance and security.
It would have been extremely costly for Microsoft to develop an alternative to the MPI technology, which "is a complex piece of software that would take years to develop," Faenov said.
But Systems Manager Robert was skeptical of the move, saying that Microsofts history in the computing arena suggests it will still try to create some type of lock-in down the road.