We have instances of stored procedures that regularly take .01 - .001 seconds that occasionally take 1 second or more. As I have mentioned in several other posts, we have a utility that queries monProcessWaits every 5 seconds and prints out waits for the spids of our C++ applications that interact with ASE via ctlib. Usually, it prints spids that have more than .5 seconds of wait time in the given period. I noticed that I would sometimes have slow stored procedure instances per our application log files but couldn't find corresponding entries for the application in our Sybase waits utility log.
I changed the utility so that for some applications it would print out their wait data every 5 seconds as long as they had at least a single wait in the current period. Still I notice that sometimes applications have slow running stored procs but do not have any waits in the period when the stored proc is slow. My question is how can this be? Put a different way, if a stored procedure that normally runs very quickly is slow in some instances and the process running the stored proc shows no wait data at that time then what is making the stored proc slow?