lkml.org 
[lkml]   [1999]   [May]   [6]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
 
Messages in this thread
/
From
SubjectRe: Ken Thompson interview in IEEE Computer magazine (fwd)
Date
> On Thu, May 06, 1999 at 12:49:54PM +0200, Pavel Machek wrote:
>
> > > Banks put your bank accounts on IBM mainframes, on HP boxes, on Solaris
> > > boxes, maybe (but unlikely) on DEC Alpha boxes. In roughly that order.
> > > Ditto for airline reservations, power grids, etc.

A lot of places also use Sequent machines.

> > We have quite a few solaris boxes in school, and they are not too
> > stable. Our ss1000 server was so bad that we changed to _experimental
> > version of sparc64linux_, and yes it improved a bit. So if you'd trust
> > solaris for your bank account, you could as well go linux.
>
> Well, banks not only use stable machines, the stability of every OS is still
> highly dependant on what you run on it. If you stress the hardware too much,
> it'll die regardless of how good the OS is designed. Also, there is this
> commonly accepted 'law of computers' that most people believe in, namely
> that you can never get rid of all bugs in software. So banks (or at least
> the rather large bank/insurance firm my girlfriend works for as a sysadmin)
> have extensive change management, at least one machine as a 'test' machine
> which they use to test changes like new software versions, and as much
> hot-swappable and self-managing parts as they can get their hands on.
>
> It's not so much 'stability' they need, it's more an issue of
> predictability. That, the rapid development of Linux, and the lack of
> support for expensive hardware such as RAID devices and hot-pluggable parts
> is what keeps Linux off the expensive-server market. IMvHO anyway.

I disagree with that. It isn't raid and it isn't hot swap. It IS high
availability. All of the companies list above have clustering products.
The idea is that you have several machines that are connected to the same
physical drives (not via NFS, but shared SCSI bus or fibre channel). In an
OLTP type environment, if one system fails, the other system can take over
its workload.

You will find that big companies that have big databases need 24/7 uptime
even if one system fails.

Even more important in a cluster is the ability to do a rolling upgrade.
Shut one system down while the other one stays up. Upgrade the OS on the
system that you shutdown and reboot. Do the same thing on the other system.
It's (in theory) the no-downtime OS upgrade. Can Linux or Windows do that?
I know that Solaris, Digital UNIX, and Sequent PTX can.

--
"I used to hang out by the food table at parties
because you don't really have to talk to anybody,
and if you do, you can talk about the food."
-- Jennifer Jason Leigh http://www.cs.pdx.edu/~idr/

-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.rutgers.edu
Please read the FAQ at http://www.tux.org/lkml/

\
 
 \ /
  Last update: 2005-03-22 13:51    [W:0.263 / U:0.468 seconds]
©2003-2020 Jasper Spaans|hosted at Digital Ocean and TransIP|Read the blog|Advertise on this site