This content has been marked as final. Show 1 reply
Actually the MIN parameter in the *SERVERS section defines how many copies (processes) of the server Tuxedo will boot by default. MINDISPATCHTHREADS and MAXDISPATCHTHREADS determines the number of threads a multi-threaded Tuxedo server will use.
As to determine if there enough copies or threads, generally you need to determine the maximum amount of concurrency your hardware can handle. So the normal procedure is to use a load test and increase the load until queues start to build up for the servers. This can be observed in tmadmin with the psr or pq commands. Once the work queued starts to increaser, check to see if there is a bottleneck resource (typically CPU). If not, then you need to increase the number of copies or threads (or both.) Keep doing that until you reach a system bottleneck such as CPU, database I/O, disk I/O, etc. At that point, you know you have sufficient servers or threads to be able to fully utilize the hardware resources available.
For MSSQ, you can still enable load balancing, LDBAL, although if there is only a single MSSQ set, there won't be any load balancing to do. Still I usually recommend enabling load balancing so if you decide to add additional servers or MSSQ sets, Tuxedo will load balance the requests across them.
Oracle Tuxedo Chief Architect