[Info-vax] Uptime for OpenVMS

[Paul Sture]

In article <4ddb22cf$0$19238$c3e8da3$a9097924 at news.astraweb.com>,
 JF Mezei <jfmezei.spamnot at vaxination.ca> wrote:

> Mark Daniel wrote:
> 
> >> 1848 Days in the PAST is Monday - May 1st, 2006
> > 
> > This is an uncommon testament, not only to the system hardware, O/S and 
> > applications but data centre design and implementation, operations staff 
> > and company management.  Thanks Keith.
> 
> 
> But also a testament of a stale system not getting new software or OS
> updates.

If it ain't broke don't fix it...

> In this day an age, it shouldn't be an individual node's uptime that
> counts, but rather the "service availability". If you run a web server
> farm designed so that you can shutdown one node at a time without
> visible impact to users, then you can perform rolling upgrades.
> 
> VMS cluster maintains the cluster uptime (but can be misleading because
> it is maintained through loss of quorum when your services are all down).
> 
> Other systems may not have the equivalent of "cluster uptime" because
> availabiility is achieved via networking gear instead of at the OS level.
> 
> Back in the 1980s, system uptime was a very important metric. Less so now.

While it's nice to see long uptimes, for systems getting regular updates 
(at the system or app level) I've always preferred to do scheduled 
reboots, to check that the startup stuff works.
 
> Cosnider the microsoft web site. Always available, but they probably
> have to reboot each machine once every 10 minutes :-)

I've certainly seen bits of it down.

-- 
Paul Sture