You can run sp_sysmon both before and after tuning Adaptive Server configuration parameters to gather data for comparison. This data gives you a basis for performance tuning and lets you observe the results of configuration changes.
Use sp_sysmon when the system exhibits the behavior you want to investigate. For example, if you want to find out how the system behaves under typically loaded conditions, run sp_sysmon when conditions are normal and typically loaded.
In this case, it would not make sense to run sp_sysmon for 10 minutes starting at 7:00 p.m., before the batch jobs begin and after most of the day’s OLTP users have left the site. Instead, it would be best to run sp_sysmon both during the normal OLTP load and during batch jobs.
In many tests, it is best to start the applications, and then start sp_sysmon when the caches have had a chance to reach a steady state. If you are trying to measure capacity, be sure that the amount of work you give the server keeps it busy for the duration of the test.
Many of the statistics, especially those that measure data per second, can look extremely low if the server is idle during part of the sample interval.
In general, sp_sysmon produces valuable information when you use it:
Before and after cache or pool configuration changes
Before and after certain sp_configure changes
Before and after the addition of new queries to your application mix
Before and after an increase or decrease in the number of Adaptive Server engines
When adding new disk devices and assigning objects to them
During peak periods, to look for contention or bottlenecks
During stress tests to evaluate an Adaptive Server configuration for a maximum expected application load
When performance seems slow or behaves abnormally
It can also help with micro-level understanding of certain queries or applications during development. Some examples are:
Working with indexes and updates to see if certain updates reported as deferred_varcol are resulting direct vs. deferred updates
Checking caching behavior of particular queries or a mix of queries
Tuning the parameters and cache configuration for parallel index creation
