[geeks] DST hell

[Jonathan C. Patschke]

On Mon, 12 Mar 2007, Sandwich Maker wrote:

> back in the mid/late '90s when i was shopping for a new isp, several
> told me that their windows servers wouldn't go a week without
> crashing.  only one was proactive, scheduling twice-weekly preemptive
> reboots.

When I was still in the ISP business, I visited one of our upstreams,
and they gave me the datacenter[0] tour and all that.  While I was in
there, I saw a guy in workclothes come in from the parking lot, through
the lobby, into the computer room, walk up to a WindowsNT RAS system,
quickly reboot it, walk out to his truck, and drive off.

I asked them what that was all about.

   "Oh, that's his job."

   "He's a sysadmin?  But he didn't even see that the machine came back
    up, and there were still users connected."

   "No, he drives from site to site every day rebooting one or two
    Windows boxes in each location.  Over the course of a week, he gets
    to all of them.  That way they're less likely to crash unexpectedly."

   "*boggle*  I don't have the budget for an extra PFY like that.  I'm
    glad I run Solaris."

Why anyone would use RAS when Portmaster 2e30s and Portmaster 3 terminal
servers worked -so- much better[1] was something I never quite figured
out.  We had half a rack of the things.  That plus a couple of RADIUS
servers and some OSPF fun, and we were cooking with gas.

Ah, the [good?] old days.


[0] "Such a -cuuute- widdle datacenter with its four teeny racks!"
[1] And, really, weren't that much more expensive, since you could make
     up a lot of the costs through offering add-on services like
     "shotgun" modem support and through buying the lines in bulk on PRIs
     instead of individually.
-- 
Jonathan Patschke ) "I would buy a Mac today if I was not working at
Elgin, TX        (   Microsoft."      --Jim Allchin, VP of Platforms