12 Jan
2001
12 Jan
'01
8:02 a.m.
Both Windows and the popular varieties of LINUX are
pretty easy to install
these days. However, Windows suffers greatly from the fact that it has
tried to maintain the usefulness of those applications that were written for
and bought concurrently with much earlier versions of the OS. My brief
encounter with UNIX systems (SUN) taught me that in order to have a
functional system, the system has to be rebuilt every time there's a new
module installed, AND all the software needs to be revised. That means
about 6 weeks of downtime every time anything is revised, be it a serial
port driver or an application patch. To have the system "up" most of the
time, it required a man on duty 24/7. That meant a burn rate of $100K per
year just to keep the OS working. I was REALLY grateful that I could unload
that piece of junk onto my client. Do you think the average "user" wants
that sort of overhead for his computer that he uses to download the latest
nudies?
>
> -Dave McGuire
Boy... that's pretty surprising since the SUN Ultra10's here are running
5 year old Solaris 2.6 (with the y2k and recommended patches from Sun).
Was this back in the Sun 3 days with SunOS under v4.1?
I've never seen this in 13 years of Sys Admin and 20 in the business.
My stuff is up over 6 months at a time (usually rebooted for lab moves,
ups battery problems, and power failures.
I've got 6 months on an old FreeBSD box which I decommissioned.
Would've been at least two years if I left it running in the corner.
I've got Sparc's running SunOS 4.1.4 with the Y2k patches.
The uptime is pretty good.
Time to patch to was a couple of hours per Sun. The Solaris boxes
patch right up with a shell script and the only downtime involved is the
5 minutes of reboot. I patch 'em all about twice per year when things
are slow (between Christmas and New Years).
I do them all in a couple of hours... from one window to each desktop.
Type reboot when done and it's finished.
The SunOS y2k was done manually -- per patch scripts run and then
the changed files were moved to a tar file which dropped on top of each
machine at single user. Machine specific tweaks were applied
and they were rebooted -- time 15 minutes/Sun once per machine and about
3 hours testing and verifying the Sun patches.
Not quite as nice as RT11's PATCH and SIPP on PDP11's or VMSUPDATE... but OK.
--Bill
--
bpechter(a)monmouth.com | FreeBSD since 1.0.2, Linux since 0.99.10
Brainbench MVP | Unix Sys Admin since Sys V/BSD 4.2
Unix Sys.Admin. | Windows System Administration: "Magical Misery Tour"