Michael Merritt wrote:
>
> lunaslide wrote:
> >
> > Michael Merritt wrote:
> > >
> > > Karsten Johansson wrote:
> > > >
> > > > Ewan Dunbar wrote:
> > > > >
> > > > > On Mon, 12 Apr 1999, Fred A. Miller wrote:
<rm>
> > Also, clustering does not equate to failover support. True, you can
> > remove a node and have the cluster continue to function, however,
> > clusters use a API that allows for custom programs to do highly
> > specialized tasks. They are not ready to support stuff like Oracle DBs
> > running over several machines (that's *next* week ;-)). Failover is a
> > totally different thing, and NT has it better than linux, I'm afraid.
>
> Show me a failover-capable NT server. I know you can't fully acheive
> this on Linux (there are several systems in the works), but if you call
> Microsoft's clustering technology failover-capable, you've never used
> it.
We use over thirty of these beasts at work in our department alone for
post offices. Two machines, same software; monitoring each other for
failures using NT's features with a little help from Compaq's server
health software. This isn't "clustering", BTW. You are mixing terms.
We also have a SunOS setup for failover in this fashion (and it is far
more robust), and it is not clustering either. Another group uses them
for network telephony applications.
I didn't say it's *great* or even good, but it's there. The worst thing
we can do as a community is neglect the shortcomings of our OS. That is
what will bring down NT. There was an interesting article a few weeks
ago on some of the shortcomings of Linux here
http://linuxtoday.com/stories/4659.html
Most important about that article was the statements about avoinding
groupthink. We can't be afraid of pointing to something in Linux and
saying "that just sucks" or "where is that feature?" If we do, we
become like the Mac faithful (I'm recovering), who do not seem to
percieve the major limitations of their OS. What we need is the same
sort of article, but more in depth, about Linux's enterprise server
limitations.
> > Finally, there seem to be efforts on making a journaling filesystem for
> > Linux, but I see no documentation on such a thing being part of the
> > default system. As far as I know, NT doesn't have one, or if it does,
> > it's capabilities are limited. An example of an OS that does is BeOS.
> > AIX is another.
>
> I know Linux doesn't have this yet... Neither does NT.
That's what I said.
> > > Also, my server does quite well using its dual processors -- I'm not
> > > sure how much more support is needed to be "full-fledged?"
> >
> > How much load does your server handle? How many open file handles, user
> > requests per second, and so on? Linux is getting there, but some of the
> > commercial UNIXs can handle stuff that Linux cannot yet. I don't think
> > that NT can, though. There's just not enough documentation out there to
> > say either way.
>
> My definition of full-fledged is it does what its supposed too... :-)
For that to apply to mission critical, you have to finish the sentence
with "under high load".
--
lunaslide * PGP key->pgpkeys.mit.edu port 11371
* * * * * *
Yep. I have a very active imagination; it's been a personal
condition of mine for years now. * -Bruce Sterling
* * * * *
--
To get out of this list, please send email to [EMAIL PROTECTED] with
this text in its body: unsubscribe suse-linux-e
Check out the SuSE-FAQ at http://www.suse.com/Support/Doku/FAQ/ and the
archive at http://www.suse.com/Mailinglists/suse-linux-e/index.html
