HP3000-L Archives

April 1996, Week 2

HP3000-L@RAVEN.UTC.EDU

Options: Use Monospaced Font
Show Text Part by Default
Show All Mail Headers

Message: [<< First] [< Prev] [Next >] [Last >>]
Topic: [<< First] [< Prev] [Next >] [Last >>]
Author: [<< First] [< Prev] [Next >] [Last >>]

Print Reply
Subject:
From:
"<Elbert E Silbaugh>" <[log in to unmask]>
Reply To:
Date:
Fri, 12 Apr 1996 12:26:46 -0700
Content-Type:
text/plain
Parts/Attachments:
text/plain (39 lines)
All responses to this logging performance stuff have been good - so here
is another (hopefully) good one:
 
This is one of the few things for which you can easily test
performance issues IN YOUR environment without a lot of effort.
In fact, it takes about 5 minutes to get it going. But you might want
to practice with your test db if you have never done this thing.
 
(1) GETLOG yourlogid;LOG=yourlogfilename001,DISC;AUTO
(2) BUILD yourlogfilename001,DISC
(3) LOG yourlogid,START
 
(3a) Make sure users have LG capability. (Though I think this is now
     not required on some newer versions of MPE/iX - I do not remember
     when version it is/was/will no longer be required)
 
You can do the above in advance.
Find a time some night when there is no access to the
databases, then do in DBUTIL:
 
(4) SET db LOGID ...
(5) ENA db FOR LOGGING
 
and monitor the day's performance - and user complaints. The next night:
 
(6) DIS db FOR LOGGING
 
If things are cool, then get down to business and think out and set up
your automagic log cycles, etc.
If performance is not cool, then dont do it.
To save breath by the purists who read this message, I did not include
DB backup process in this because it is not required to merely test
performance issues.
 
Outside of when we used to have a Series III machine, I have never been
able to measure any CPU requirements above several % points.
 
Elbert

ATOM RSS1 RSS2