% fortune -ae paul murphy

The deed is done: now are you better off?

Scenario 1:

Congratulations! The virtual "you" has done the deed: the last HP RP8440 - always a relative piece of junk because designed for Itanium and sold with the incompatible PA-8900 - is gone, and a couple of nice shiny new Proliants have taken its place. Linux rules! HP-UX? is history - and you've got a few thousand bucks left over after "cashing in" your support budget on the capital change.

This is a clear win because you got big gains out of it: the Proliants are faster, cost less to buy, less to maintain, will likely have comparable or better reliability, and support a far wider range of high quality software - much of it free and all of it cheaper to use than any PA-RISC counterpart it may have.

The only downside is that now you're a little more susceptible to both external (PC security style) attack and internal (Wintel monomania) attacks - but hey, for you, neither is a big deal.

Scenario 2:

Congratulations! The virtual "you" we've been imagining has done the deed: the last Sun SPARC box is gone, and a couple of nice shiny new Proliants have taken its place. Linux rules! Solaris? is history - and you've got a few thousand bucks left over after "cashing in" your support budget on the capital change.

But now I can't copy the second paragraph above for use here because the benefits aren't so clear: Yes, the Proliants cost less now than the old SPARC machines did - and they'll cost less to maintain - but an upgrade to the T2 line would have cost even less; the Proliants aren't really faster overall but a T2 would have been; the Proliants are neither quite as reliable nor quite as secure as the old gear; you've lost Solaris capabilities from dTrace to ZFS; and, any Linux software you can run on them could also have been migrated to both old and new SPARC machines with very little trouble.

Still, you're arguably better off because.. What? lots of people who don't use either one praise Linux over Solaris?

Given some specifics I could probably argue this one either way - but think that there are more cases in which the upgrade to a T2 (or an M series) would have cost less, foreclosed fewer options, and been strategically as well as technically preferable, then the other way around.

The difference I think is this:

• going from HP-UX with PA-RISC in an Itanium box to Linux on a higher end x86 machine is a sidestep out of a deadend and back onto the road, because HP has all but abandoned both to concentrate on putting bruises on other people's bananas - and, of course, on a personal basis, the fact that the company has never had the commitment to building customer careers that makes IBM the real employer of reference for most IBM loyalists, means that there's really no reason not to abandon them.

• going from SPARC/Solaris to Linux on x86, in contrast, is a clear leap backward on both hardware and software because there isn't much you can do on Linux that you can't do better on Solaris, while Sun's "coolthreads" stuff is leading edge -an order of magnitude better on bang for the buck than x86.

And that's really the bottom line difference: both scenarios usually work on the money, but one makes strategic, business, personal, and technical sense - and the other often doesn't.