Re: Sun may have Solaris for Itanium already ready
From: Chris Morgan (cm_at_mihalis.net)
Date: 30 Oct 2003 10:27:23 -0500
Gavin Maltby <G_a_v_i_n.M_a_l_t_b_y@sun.com> writes:
> > I don't think this issue is completely resolved yet, but that happened
> > more than 18 months ago now - eons in Linux time.
> So it's probably only just beginning to be trustworthy enough to
> commit your enterprise data to, unless it saw masses of testing
> before being integrated (and the 24 hour suggestion wouldn't
> suggest that). Stable Linux releases would only be including
> the feature now'ish. But bleeding edge or interim releases
> would have been chancing their data to luck for a long time.
redhat kept developing the previous VM for precisely this reason,
nevertheless Linus put the new one right into the 2.4 kernel because
it was 'better' in his view right away. The other implementation team
then complained that he was ripping out something which hadn't
received their recent bugfixes to replace it with somethint else that
was fresh the developer's hard disk.
> Don't get me wrong, I'm not really dissing this development you
> mention (because I don't know the real details). I'm just
> saying that VM system bugs can silently corrupt your business data
> in the blink of an eye, so give any redesign a long soak time
> before you accept it into production. With vendors this time
> has mostly come in the longwinded OS release cycle - features
> may not appear overnight but when they do they are usually
> well up to scratch.
It's true that you would traditionally want a long period of time to
elapse after changing the VM before you rely on it in
production. Without that, however, the alternative is intensive
distributed testing over a huge diverse community of machines. Linux
has that better than most other operating systems
-- Chris Morgan "Post posting of policy changes by the boss will result in real rule revisions that are irreversible" - anonymous correspondent