Disabling Monitors

[Mark]

If I understand correctly, by default, simmer records data about different processes using a similar approach to the old "monitors" in SimPy.

Is there an option to disable monitoring?  

Also, SimPy had an alternative approach called "Tallies" that only kept a running aggregate value (such as mean, stdev, count) for the purpose of keeping memory consumption small.

For very large, or very long-running simulations, what is the best way to only collect statistics on the metrics of interest?

[Iñaki Úcar]

2018-03-12 20:26 GMT+01:00 Mark <slash...@gmail.com>:
> If I understand correctly, by default, simmer records data about different
> processes using a similar approach to the old "monitors" in SimPy.
> Is there an option to disable monitoring?

Yes, there is. Every resource, and every arrival source, has a "mon"
argument. If it's set to FALSE, that particular resource, or the
arrivals generated by that source, are not monitored.

> Also, SimPy had an alternative approach called "Tallies" that only kept a
> running aggregate value (such as mean, stdev, count) for the purpose of
> keeping memory consumption small.
> For very large, or very long-running simulations, what is the best way to
> only collect statistics on the metrics of interest?

The best way is to disable what you don't need. If it still does not
fit, you need to purchase more memory (or you have to rethink your
simulation ;-) ).

It's almost impossible to fit all the possible needs by implementing
some running summaries. And it's a burden to have to think about what
you have to collect and where. So the *simmer way* is to collect
everything automatically. Every decision has advantages and
shortcomings. This is *our* approach.

Nonetheless, there are some improvements to come. The implementation
of some kind of "connector" is on the roadmap, so that we will be able
to plug simmer's monitors to, e.g., files on disk or a SQL database.
But that's a medium-term goal.

I hope I have responded.

Iñaki