Artem
ALS add a new SPID in ASE, what we seen is that the cost of
executing the new SPID is high, it requires one engine to be dedicated.
So we have seen delays in this case below.
* Low logging activity, there is a delay which may cause high
transaction response time.
* Unchained mode transaction delay which may cause high
transaction response time.
We do not have any performances data to look at here but I
would think this is related to how the ALS spid is behaving.
So if workload is not very high for the engine up against
the 90% User busy for long periods of time and you don't see log contention for the sane long
period of time which is not normal for a production system, I would not recommend to configure this one.
I would check a few other things first to improve the logging.
1) What is the response time for the syslogs device at the
time of that this is seen. (from the OS and ASE quick writes reduce contention)
2) Could this database syslogs be bound to it own cache with
a type of "Log Only". (You must bind a database, user table, index,
text object, or image object to a cache of type “Mixed.” Only the syslogs
table can be bound to a cache of type “Log Only.”)
3) Check that "user log cache size" have the correct sizing.
hopefully it give some help to improve the performances.
Niclas