Solid technology helped sustain OpenVMS. One of its chief advantages is strong "clustering" capability, in which two or more systems are linked together so one can take over if another fails. "Even today, it's probably better at clustering than most environments out there," Haff said. "OpenVMS' cachet is rock-solid reliability." The quality of the clustering has meant customers could switch off a system, install new hardware and fire it back up again, all without the software skipping a beat, Haff said. "We think... somewhere between 90 percent and 95 percent" of OpenVMS customers use its clustering features, Gorham said. "It has been one of the signature areas." HP will support OpenVMS clusters built of both Alpha- and Itanium-based computers, an essential step to making the migration. One challenge for OpenVMS in the future is staving off ever-improving systems that have more mainstream use: Windows, Unix and Linux. When computers ship in higher volumes, component costs decrease, software companies are more interested, developers are easier to find and support is easier to come by. "Any time you design something that's general purpose," Haff said, "it will rarely be as good as the specialist. But over time in the computing industry, general purpose usually wins. General purpose usually becomes good enough." Where Unix edged in on the VMS turf once it became good enough, Linux now is edging in on Unix. And Digital engineers are helping that happen. Clustering programmers from Digital, including Brian Stevens and Tim Burke, moved to a start-up called Mission Critical Linux and now are working at top Linux seller Red Hat. That clustering work now is part of Red Hat Advanced Server, said Stevens, now vice president of operating systems at Red Hat. VMS was designed to support clustering from the ground up, but it's harder to retrofit such deep changes to an operating system, Haff said. Digital was gradually building clustering into its own version of Unix, improvements that eventually will make their way into HP's version, called HP-UX. HP agrees that it's foolhardy to try to deny mainstream computing trends. Indeed, its OpenVMS strategy is largely a move to embrace the mainstream technology as much as possible, a move that began when VMS was renamed OpenVMS after it acquired the ability to run some Unix software. More recently, OpenVMS was fitted with the ability to run Java programs. And for the latest technology, HP is bringing a version of the open-source Apache Software Foundation Web server software to OpenVMS, Gorham said. The move to Itanium will keep OpenVMS closer to mainstream hardware, as well. The Itanium move means HP's computer development cost will result in a single line of products that can run OpenVMS, Linux, Windows and HP-UX. (Separate Itanium hardware will run HP's fifth operating system, the ultra-high-end NonStop Kernel.) By mid-2004, HP plans to sell OpenVMS servers with two, four, eight and possibly 16 processors, Gorham said. It plans 32-processor and 64-processor versions by late 2004, and, if all goes well, 128-processor Superdome support in 2005. With the unified hardware foundation, HP believes it will be able to keep its OpenVMS competitive. "The strategy for HP is to take advantage as much as possible of the standardised world," Gorham said.
Is Linux really a viable alternative to Windows on the desktop? Read the latest headlines at ZDNet UK's Operating Systems News Section.
Let the editors know what you think in the Mailroom.
