Lost in the excess of press and collective angst over the fate of HP’s HP-UX servers and the widely-accepted premise that Itanium is nearing the end of its less than stellar run has been the fate of HP’s NonStop users. These customers, some dating back to the original Tandem customer roster, almost universally use HP NonStop systems as mission-critical hubs for their business in industries as diverse as securities trading, public safety and retail sales. NonStop is far more difficult to engineer out of an organization than is HP-UX since there are few viable alternatives at any reasonable cost to replace the combination of scalable processing power and fault-tolerance that the NonStop environment provides.

NonStop users can now breathe collective sigh of relief – on November 4 HP announced that it was undertaking to migrate NonStop to an x86 system platform. Despite the lack of any specifics on system details, timing or pretty much anything else, I think that NonStop users can take this to the bank, figuratively and literally, for a couple of reasons:

  • HP has a pretty good track record of actually delivering major initiatives that it commits to. Their major stumbles in dealing with their Itanium-based HP-UX program has been in not communicating rather than missing commitments. Technically, given another cycle of server CPUs and their collective expertise in systems design, including the already undeway high-end x86 systems programs, there is little doubt that HP can deliver a platform suitable for supporting NonStop.
  • NonStop’s fault-tolerant internals are less dependent on tightly couple hardware than are other alternatives than are other alternatives, such as Stratus. They have also been successfully ported through three generations of hardware (Tandem, HP PA-RISC and HP Itanium), so it’s a pretty good bet that the code base is already structured for easy migration.
  • It’s not a charity – the NonStop base represents a perpetual annuity with high margins for HP. While HP talks about introducing new technology that lowers the cost of NonStop, they also realize that the costs of switching for NonStop users is high, and they can price the systems at a premium to equivalent high-end Linux or Windows systems. This is not to say that the systems will not deliver high value – they will, and almost certainly more cheaply than a cluster of conventional systems engineered to deliver the same reliability. BUt there is little pressure for HP to price them as if they were a commodity system ina competitive market. Accompanying services will also carry hefty margins.

All in all, NonStop customers should find this announcement reassuring, and can expect that with approximately 24 months they will have a viable alternative that effectively guarantees the continuation of what has become the lynchpin of their enterprise systems of record.