Re: High availability questions -- MIDRANGE-L

On Fri, 23 Mar 2001 12:50:06 -0600, "Stone, Brad V (TC)"
<bvstone@taylorcorp.com> wrote:
>Expensive and not 100% reliable.  Remember, you have a network as part of
>the solution, and these products don't help your network stay up.  In fact,
>there's usually a lot of traffic which can hurt the network.  

That's why we have a dedicated fiber optic line between our two 640's.
Nothing but the mirroring software uses it.

>Or, the network goes down for a few seconds, your replication product may
>stop.

Yes, but what are the odds that at that same instant the production
AS/400 goes down?  In the three years we've had HSA, I believe our
FDDI line has went down once.  Our production AS/400 failed once.  The
two events were many months apart.

>My advice (honest here, too).  Have them buy a new AS/400.  Take it home.
>Next, take your current production AS/400, add another IP address and tell
>folks that's the new machine.  Even better, set up another partition on the
>machine.  Save while active, and you're done.  :)  

Save while active is a cool feature, but it doesn't equal the ability
to do a full production backup from a mirrored system.  Plus, the
first time you lose the primary system and you're not back up within
an hour, someone's head will be rolling.  At least that's what would
happen here after buying a second AS/400 for HSA and not really having
it.  Yes, I know you were just kidding.  :-)

>If that doesn't help, I guess we'd need to know more what type of
>replication you'd be doing.  ie just data, or all objects.  And, if one goes
>down, would you want it to automatiaclly "switch" over the other machine?

Our software could switch over automatically, but we choose not to.
If the primary machine fails we'll know about it within minutes.  We
can then make a judgement call at that time as to whether to come up
on the backup machine.

>Well, there are flaws with those theories as well.  We tried it, and finally
>decided that it was more trouble to try and replicate two AS/400s (at the
>object level) than to just not worry about our machine going down.  If it
>was a total catastrophy then it would take a max of 2 hours to load up
>another AS/400 to take it's place.  That's not too much to ask.  24/7 is a
>want, not a need.  Otherwise everyone would be out of business.  (ie they

Good luck on the two hours.  When our F85 went down before we had HSA,
it took IBM almost a day to get the parts in here that were needed.
We wound up being down almost 36 hours.  It was bad.  Very, very bad.

>were never 24/7 before it became a buzzword).  Not to mention, if your
>business can't survive with a couple of hours of downtime, maybe it should
>be re-evaluated if it's viable.  They say you should be able to run your
>shop for 6 months on no income.  :)  

It could be more than just a few hours of downtime, depending on the
situation.  Ask yourself what it would be like being down a day or two
when thinking about HSA.  I can tell you right now our company would
not be able to operate for six months without the AS/400, I wouldn't
even want to see what it would be like for a week.  36 hours was bad
enough!  So bad the administration had no problem budgeting for a
second AS/400, HSA software, and many man hours to get mirroring set
up and running, as well as maintenance and on call support to keep
things going.  Is it worth it?  YES!

Terry Herrin
Sr. Programmer/Analyst
New Hanover Regional Medical Center
Wilmington, NC
+---
| This is the Midrange System Mailing List!
| To submit a new message, send your mail to MIDRANGE-L@midrange.com.
| To subscribe to this list send email to MIDRANGE-L-SUB@midrange.com.
| To unsubscribe from this list send email to MIDRANGE-L-UNSUB@midrange.com.
| Questions should be directed to the list owner/operator: david@midrange.com
+---