<<< HUMANE::DISK$SCSI:[NOTES$LIBRARY]DIGITAL.NOTE;1 >>> -< The Digital way of working >- ================================================================================ Note 5344.11 Looking for clustering info. 11 of 11 STAR::KLEINSORGE "Fred Kleinsorge, OpenVMS Engineer" 30 lines 19-JUN-1997 11:32 -------------------------------------------------------------------------------- .9 don't confuse the Galaxy architecture with specific platform features. Hot-swap is a hardware feature not-yet-available on existing platforms. Galaxy finesses some of the really hard problems with so-called hot-swap or in-swap. Hot-swap by some O/S's is the idea that dual path devices (like disks and networks) can fail over when one of the paths is removed (powered down). Galaxies can finesse the problem by shutting down a portion of the system. Of course, when PCI per-slot hot-swap arrives, there is *real* work needed to handle that thorny problem. .10 Yes. I think that Galaxies can be a pull for application vendors. If they want to clain to be the fastest (and we hope by a large margin), then they will need to be able to run on the box. The box itself can be thought of by the *consumer* as a black-box application engine. We are working hard with some key ISV's to have them hit the floor running with Galaxies, and I believe that this will create the pull on the other vendors to look again. The nice thing about ISV's is that they are usually profit driven, not religious about the O/S. Servers are where the big margins are, and where volume doesn't have to be in the millions of copies to be hugely profitable. Just a note... to achieve the numbers that have been bandied about in the press and on the network for performance would require a *very* expensive system. The disk requirements alone would consume a measurable percentage of DEC's yearly disk production just to run the benchmark. So at the high-end of this market, we are talking big profits for those who can take advantage of it.