Experiment Execution

[riccardotommasini]

We recently came across with the problem of experimental workflows.
We noticed that the implemented one, i.e. warm-up, steady state, cool down,
it is just an option and it was the most common one among existing benchmarks.

Actually, they do not expand the workflow but claim they results are obtained at
steady state, which implicitly calls for a warm-up phase.
This open interesting research scenarios where we can study new workflows.
One of my doubpt was, does it exists the master workflow?
I'd like to achive a declarative way to design it, therefore I immediately though about
how to implement it.

This is just an idea.
#rsp.deploy(e).execute(15,"minutes").undeploy()

The rationale is that any experiment but be deployed, i.e. it has some initial conditions that
need to be achieved (at least engine and stream running).
Then the experiment instance depends on its definition.
The standard way
#rsp.deploy(e).execute(15,"minutes").undeploy().report()
#rsp.deploy(e).execute([events]).undeploy().report()
#rsp.deploy(e).execute([monkeys]).undeploy().report() this refers to the caos monkey testing idea

[riccardotommasini]

#rsp.deploy(e).execute([events]).undeploy().report()
#rsp.deploy(e).execute(Events.duration(15, "minutes")).undeploy().report() creates the event of duration expired

Events.add_query(q, timestamp)
Events.remove_query(q, timestamp)
Events.remove_stream(s, timestamp)
Events.stream_rate(s, speedup, timestamp)

[riccardotommasini]

rsp.deploy(e).execute(Events.duration(15, "minutes”).on_occurence(rsp.undeploy(e)).report()