The HP NonStop is part of a dying breed of high-reliability computers - a sort of last stand for the vision of the works-no-matter-what server appliance. They're unique in a lot of ways. They run an operating system (variably called NSK, Guardian, or NonStop OS) unlike pretty much anything mainstream - each processor runs its own instance of all necessary software, including the OS, and communicates with others by message-passing. This, combined with a proprietary cluster interconnect, lets HP claim that NonStop scales "near-linearly" to up to 4,080 processors. Due to the fact that there are bandwidth limitations on the cluster interconnect, this isn't quite true, but NonStop applications certainly scale better than almost anything else on the market. As a result of its performance and reliability characteristics, there hasn't been an exodus from the NonStop like there has been from the mainframe and, to a lesser extent, UNIX. You'll mostly find NonStops in telecom usage (sometimes very large systems) and companies with large business intelligence or database workloads, especially in the modified stripped-down NeoView configuration. The fact that they cost several hundred thousand dollars in an entry configuration means NonStops aren't expanding out of their niche, but within that niche, they do just fine.

Log in or register to write something here or to contact authors.