ORBIT-USER: How do you define an experiment

[Max Ott]

Folks,

I'm interested to know how each of you arrive at the definition of the
experiments you run? How do you arrive at the choice of resources
(nodes), the parameter settings, the kind and source of traffic, and
the sequencing of parameters throughout a single experiment, as well
an entire investigation. What's the process of translating the type of
analysis you want to perform to an actual experiment description?

The reason for this is that the entire process right now requires us
to first identify all the resources we need and then try to obtain it.
As we are dealing with (highy) unreliable resources in a real testbed
we often get stuck.

What I'm interested in, is to find out if we can capture the intent
(or characteristics) of a set of an experiment and the associated
resources and then leave it to something else to find the appropriate
mapping to resources based on what is available.

A concrete example would be topology. Right now we are dealing with
node x at y. How did we choose x and y? Can we instead talk about 'node
density', in/out degrees, ...?

Any feedback would be highly appreciated.

Thanks,

-max