Using only statistics available within the database, this calculation gives the best approximation of response time, in seconds, per transaction during a sample period. This is useful in further diagnosing performance issues by drilling down into individual components like Waits, CPU etc.
I've looked at your article and wrote a pl/sql process to track the response time for metrics. We have a very badly written application that we are supporting.
Question: How do you handle the negative results? Am I looking at this correctly?
Hi guys,
I know this is an old thread but what I require is related to this.
I have used the following script to get transaction speed on my DB based on the formular supplied by UDAY but the results are unclear. Can you tell me if this is really the number of seconds used per transaction or is my script incorrect:-
select ((WAIT + val) / TRXN_CNT) as TRXN_SPEED
from (select sum(TOTAL_WAITS) as WAIT from v$system_event) ,
(select value as VAL from v$sysstat where name='CPU used by this session
'),
(select sum(value) as TRXN_CNT from v$sysstat where name in ('user rollb
acks','user commits'))
/ SQL> /
Bookmarks