I am running a 64 bit ASE on solaris 2.8 cpu's- 8;
since upgrade from 11.9 ..

1,Over all performance for queries that return large result set has
been better than in 11.9.

2, However for queries returning few rows and few columns, there has
been a 3 TIMES slower, during loaded condition,
1, Concede..64 bit would slow these donw may be 30%
2, part of the problem is poor coding (legacy) that loop thru
200,000 calls to the DB , instead of one call and cache the result
set.
3, This behavior is consistent for batch jobs as well as web based
application( thru jconnect)
So, besides spending eternity fixing all the years of junk code.
Trying to figure out why?


Looked thru sysmon/wait time / etc..etc.. between loaded and
unloaded server.. could not find anything wrong. Now as a last resort
considering
1, adding a new network listener to see if this makes a difference
2, separate tempdb cache for this application
Question:
Any comments on these ventures..?
Has anyone seen this behavior on 64bit - ASE 12.5.1 on solaris

SECOND PART

Following EBF migrated
Adaptive Server Enterprise/12.5.1/EBF 11658 ESD#2/P/Sun_svr4/OS
5.8/ase1251/1838/64-bit/FBO/Fri Feb 20 12:35:43 2004


Why did we apply this EBF..?
1, Bug in Group by ..with a variable name was return duplicate rows
2, Sybase advisory on potential of "Urgent from Sybase: ALTER TABLE on
DOL Table With Placement Index May Cause Data Loss in ASE 12.0.x "


Issues since then..
1, Zombie process were found.. involved a server recycle

2, Server stopped accepting packet size greater than 512 bytes .. So
rolled back this EBF for something more stabler.

No sybase response on fixes, since probably we were one of the first
to try this on 64 bit.. Anyone else seen this behavior..

Sanity check..

regards

badri