[Ns-developers] emulation use cases

[Joseph Kopena]

On Thu, Jan 15, 2009 at 12:46 AM, Tom Henderson <tomh at tomh.org> wrote:
> Are there other suggestions for the above type of use case, or other use
> cases for emulation that people would like to see prioritized?

Hi Tom, Craig, everyone,

Two small notes on emulation:

- Like Gustavo mentioned, the main interest in emulation among my
group(s) is to test if protocols actually hold up in the wild.  It
would be really nice to use exactly the same code for a live test as
in a simulation, and it sounds like support for that is there or
almost there.  Integration w/ Orbit, as discussed in the past for
NS-3, would also be nice as we have people doing that as well and
eventually hope to merge their work with some other NS-3 based work
here.  If the module can run live traffic then this might be a
non-issue.

- Emulation lets you test practical computational complexity, which is
very difficult to do in simulation.  For example, if a routing
protocol's core logic is too slow to keep up with load, that will show
up in live and distributed emulation tests.  It won't show up in
simulation because there is essentially infinite processing time
between events.  I'm aware of one or two (essentially homebrew)
simulators that try to address this, but they seem messy to work with
and the results in this regard questionable.  This could be a notable
consideration for people doing work on deep packet inspection, content
based routing, or application layer overlays.  It came up a few weeks
ago in my lab w/ a couple guys writing database driven declarative
routing systems in NS-3, though we decided in the end our processing
load was not high enough for it to be a real issue at the moment.

Thx

-- 
- joe kopena
right here and now