RE: Is this Needed?

I'm not going to make a stink about this. Its expected to take about two hours and it's not my system. However, if the technical team lead should suggest that *I* do this in order to improve my performance then I'll have a few things to say. The politics of the situation are that I generally respect the other DBA's skills. He (a contract DBA) knows a lot, and has many more years Oracle experience than me But doesn't generally feel obligated to explain what/or why he is doing things. He doesn't work for me, nor does the technical team lead (another contractor).  

I am a full-timer and have pretty much open-door access to anybody I want to talk to up the chain if I see something fishy. I've only had to do that one or two times for the word to get out that it isn't safe to ignore my concerns. In this case, I don't see any harm arising from the effort although, some correspondents here have pointed out that it is possible that performance will decline. This isn't, by the way, the first time they have done this and I think the primary reason is to reduce fragmentation and improve scattered reads. Since the disks are RAID 10 I'm not sure there will be any measurable improvement and haven't heard that any customers are complaining. My perception is that the response time is slowing somewhat and that the long-term solution is to archive a lot of older data.  

Don Freeman
Database Administrator 1
Bureau of Information Technology
Pennsylvania Department of Health
717-783-8095 Ext 337

-----Original Message-----

From: ryan_gaffuri_at_comcast.net [mailto:ryan_gaffuri_at_comcast.net] Sent: Wednesday, October 26, 2005 12:22 PM To: dofreeman_at_state.pa.us; Oracle-L (E-mail) Cc: Freeman, Donald
Subject: Re: Is this Needed?

if this is a major effort, ask them to show a proof of concept that this will improve performance. Ask for tangibles results. For example, the response time for this screen during the day at peak usage is 'X', doing this will reduce response time by 'Y'.  

Or it will reduce CPU usage(and you really need to reduce CPU usage because you need to add more usage) and have him prove it to you.  

Logic doesn't matter. Make them prove it. You can always test performance. Have them transport a production tablespace to another instance and run tests. If you do not see any tangible results(not just reduced numbers from readings in the database, but actual improvement in performance) its not worth the effort.  

Typically putting up this kind of roadblock kills silly ideas, since the people who come up with them don't want to work that hard to prove to you that they will do anything.  

--

http://www.freelists.org/webpage/oracle-l Received on Wed Oct 26 2005 - 12:23:40 CDT