Re: Why would one queery hav 20x more sql*net message from client[idel] than other almost idenditcal query

In article <1172776286.910301.21290_at_k78g2000cwa.googlegroups.com>, "Valentin Minzatu" <valentinminzatu_at_yahoo.com> wrote:
>On Mar 1, 9:35 am, mistton <s..._at_dsfds.com> wrote:
>> Running Oracle Version: 9.2.0.3.0 on Solaris 8 and Sun E15 10 CPU server.
>>
>> have client app from vendor (JAVA applet on IE 6) that runs an Oracle
>> process at user request. The process is taking 20 minutes to run which is
>> totally unacceptable
>>
>> Client is just sending commands to do stuff on database it does virtually
>> no more work than that. (on client box its at 0% almost 99% of the time).
>> once Oracle is finished it displays results by reading final results table.
>>
>> looked at what oracle is doing and found that CPU, I/O, reads etc taking up
>> only 2 minutes rest is almost all idle time.
>>
>> PROBLEM:
>> One statement (almost identical to many others) is taking up 50% of SQL NET
>> idle time, and it spends 95% of its total time in it. The statement is
>> reading 1 row from a 1 row table holding currency info for USA, all
>> varchar2 or date.
>>
>> is it possible that say client does 10 queires problem one being last one
>
>
>What exactly makes you think the time is spent executing database
>code? Have you ruled out the client side completely? (at a glance, the
>stats above seem to suggest that very little of the 20 minutes is
>spent inside database)
>

that is exactly my point. the Java app server is doing alot of the work with Oracle just waiting

BUT why so much time with this one small query? is it just becuse the trace parked all the idle time there?? or is there something special that is a sign of some problem. Received on Thu Mar 01 2007 - 20:48:01 CST