Subject: unify performance problem on wednesday
fyi .
- - - - - - - - - - - - - - - - - - - - - - forwarded by brenda f herod / hou / ect on 05 / 05 / 2000
09 : 01 am - - - - - - - - - - - - - - - - - - - - - - - - - - -
rita wynne
05 / 05 / 2000 08 : 11 am
to : brenda f herod / hou / ect @ ect
cc :
subject : unify performance problem on wednesday
fyi . . . . . . . . .
- - - - - - - - - - - - - - - - - - - - - - forwarded by rita wynne / hou / ect on 05 / 05 / 2000 08 : 10 am
- - - - - - - - - - - - - - - - - - - - - - - - - - -
from : dave nommensen 05 / 04 / 2000 06 : 56 pm
to : george smith / hou / ect @ ect , edward terry / hou / ect @ ect , tricia
bowen / hou / ect @ ect , randall l gay / hou / ect @ ect , patti sullivan / hou / ect @ ect , pat
clynes / corp / enron @ enron , carlos j rodriguez / hou / ect @ ect , robert
superty / hou / ect @ ect , bryce baxter / hou / ect @ ect , rita wynne / hou / ect @ ect , donna
greif / hou / ect @ ect
cc : jeff johnson / corp / enron @ enron , tommy j yanowski / hou / ect @ ect , beth
perlman / hou / ect @ ect , scott williamson / hou / ect @ ect , jim ogg / hou / ect @ ect , regan
m smith / hou / ect @ ect , kenneth m harmon / hou / ect @ ect
subject : unify performance problem on wednesday
as you are aware , the unify system experienced significant performance
problems wednesday from approx 9 : 30 am - 11 : 20 am . the duration of the
problem was due to a type of system behavior we had not experienced before .
the problem was exacerbated by a multitude of additional user logins thinking
that a fresh login might work better . additional logins never improve
performance and generally degrade it , but that is another story . ( there was
actually a single user with 21 sessions ! ) we had to " bounce " the server
twice to bring the system back to its normal level of performance .
the good news is that we found out what caused the problem . it was not a
particular user or a particular report or process . it was actually an
intentional change to a specific table that we have done numerous times in
the past , but this particular change had drastic side effects . we have
contacted sybase with this problem to see if they can help determine if the
behavior is a " feature " or a bug . in the meanwhile , we will do these types
of changes at night during scheduled outages .
i apologize for the frustration , anxiety , and extra work brought on by this
problem and will take every precaution to see that it does not happen again .
if you have any further questions or comments , please feel free to give me a
call .
dave n .