Re: Less Down Time (was: Mean Time Between Failure)

[PaulMmn <PaulMmn@xxxxxxxxxxxxx>]

>  >  James Rich [james@eaerich.com] mumbled (in part):
>
>>   Can anyone comment on:
>  >
>>   1.  How to minimize downtime due to backup.  For example, how to get a
>>   good "snapshot" of the data while minimizing the time spent in a
>>   restricted state.
>>

We use MIMIX (forget the vendor company name).  We have 2 AS/400s.
The production database is mirrored to the 2nd box via MIMIX.  At the
appointed time we suspend updates to the copy, and save the copy to
tape.  Once the backup is completed, we release the hounds and let
the copy catch up with the transactions (which have been queued on
the 2nd machine).

This works well; users don't have to get off of the machine, and we
have a complete backup.  The backup is library-by-library, but since
our main application is a BIG chunk of our data, we like it.  The
users like it.

>  >  2.  How downtime due to system maintainance (daily backup, PTFs,
>>   upgrades, etc.) compares between OS/400 and other stable OSes (like
>>   Solaris, OS/390, other unix).  Does anyone run DBs on something besides
>  >  iSeries and can comment on their backup process?

Saturday night we may (read: usually) kick the users off to install
program changes.  The reason the users have to get off is if we are
installing changes to database or display files-- we need a clean
shot.  Program changes don't require users to be off.  We chose
Saturday night so all users have a chance to appreciate the changes
on Monday morning.  (:

Once a month we kick the users off to IPL the machine, reorganize
databse files, install PTFs, etc. etc.  This weekend, for example,
IBM installed a whole bunch of new disks on both of our machines.
This does require a dedicated machine!

--Paul E Musselman
PaulMmn@ix.netcom.nospam.com