Re: Oracle/OpenVMS Performance Question

In article <jkennedy-1110941346010001_at_jkennedy-lap.us.oracle.com>, jkennedy_at_oracle.com (Jef Kennedy) writes:
|>Nick,
|>
|>Kudos to you for being so thorough in your description! Seems that you
|>come from a system mgmt background, true? Do you have a DBA on site? Are
|>you the DBA, too? Let's go through some of your post...
|>

I'm sorry I missed this post, it sounds intersting.

|>In article <1994Oct10.142138.1_at_rhine.is.rice.edu>,
|>nmetro_at_rhine.is.rice.edu (Nick Metrowsky) wrote:
|>
|>> FYI, our Oracle SGA is 32 mb.
|>
|>You may want to consider raising this. More on this a bit later.
|>
|>> Our Memory Utilization
|>> averages about 60% (very rarely do we peak above 65%).
|>
|>Cool...you should determine if Oracle could make efficient use of this
|

I am working in VMS and we found a big increase in general performance by increasing the SORT_AREA_SIZE parameter in the init.ora. We found most of our sorts were 1M or less, so we set the SORT_AREA_SIZE a little over 1M and reduced the number of sorts to disk dramatically. That also improved our I/O picture.

|>> Some users report a 5 minute
|>> response time in performing a commit.
|>
|>Yikes! Depending on what your users are doing, this sounds *way* unacceptable.
|>
|

Since I missed the original post, I'm not sure if you have looked into locking activity (via SQL*DBA). Also, what else is going on when the user hits commit? Is the application validating the data behind the scenes, or are triggers updating other tables, etc.? It may not be a simple operation.

|>BTW, is this information generally useful to the rest of you NET people?
|>

Yes!

Jen

-- 
Jennifer Corliss is on an InterGalactic cruise...       |The opinions expressed 
                                                        |above may not be those
   ...in her office                                     |of my employer.