You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
One nice thing about Shadow 2.0 is that we should be able to simulate a heterogenous network where not everyone is running the exact same tor binary. It'd be nice if tornettools had support for creating such networks.
I'm not sure of the best way to specify though. Naively maybe there could be an alternative to the --tor flag that took a path and a fraction, and either could be provided multiple times or took a list. e.g. --tor-mixed=0.8:~/bin/tor-latest;0.2:~/bin/tor-old.
I could imagine wanting more sophisticated models, or maybe even basing it on real network stats, but not sure whether that's needed. Will ping mikeperry to elaborate.
The text was updated successfully, but these errors were encountered:
In the short term the immediate purpose is to test with a heterogenous mix of congestion control options; we'd be looking for a mix of clients, and they can be chosen uniformly at random. He also mentioned that it'd also work in this case to point them at different tor configs rather than using different tor binaries (and might even be preferable).
Even if we don't use it for this experiment, ahf did mention being excited to be able to use different tor binaries, including relays and exits. Apparently this has been one of the reasons to use chutney instead of shadow for testing.
mikeperry is going to spend a bit more time coming up with a concrete plan for the congestion control experiment(s) and what we'll need in the short term.
Via conversation with mikeperry -
One nice thing about Shadow 2.0 is that we should be able to simulate a heterogenous network where not everyone is running the exact same tor binary. It'd be nice if tornettools had support for creating such networks.
I'm not sure of the best way to specify though. Naively maybe there could be an alternative to the
--tor
flag that took a path and a fraction, and either could be provided multiple times or took a list. e.g.--tor-mixed=0.8:~/bin/tor-latest;0.2:~/bin/tor-old
.I could imagine wanting more sophisticated models, or maybe even basing it on real network stats, but not sure whether that's needed. Will ping mikeperry to elaborate.
The text was updated successfully, but these errors were encountered: