New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Future advanced networking improvements #815
Comments
@m-anish hope it's Ok I'm moving this to IIAB 6.7 Milestone as our IIAB 6.6 final release is now approaching? |
Cool. I hope to start experimenting on this soon. |
Update: advanced options now in play Since I'm in a writing mood other options available but less known |
I take it these have precedence over everything else, for example gui_desired_network_role == "LanController" is overruled by iiab_lan_enabled: False or nobridge: True |
'iiab_lan_enabled: False' is the big hammer to turn off all the "LAN" configuration, should effectively create an 'Appliance' like setup on the WAN side and ignore all other interfaces present. |
I'm creating this as a placeholder for possible future networking upgrades to the IIAB.
In large IIAB setups, QoS is increasingly an issue - one user watching a video eats up all the available wifi bandwidth. Something like radius can limit this problem. Another option can be packet filtering/shaping as described in the cookbook here -- http://lartc.org/howto/lartc.cookbook.ultimate-tc.html
There are many instances where upstream network connectivity is spotty (I don't need to give examples I think), there might be use cases where having more than one upstream connection can really improve conditions. Thankfully, multipath-tcp (http://multipath-tcp.org/) has been making some good strides on this front. To have a user experience that is transparent to the user and provides better and more reliable upstream connectivity, one would need to have multipath-tcp run on the IIAB (seems trivial to do on a recent debian/ubuntu machine), and have a multipath-tcp proxy hosted somewhere. The downside is that additional network bandwidth on the proxy/bridge end will be consumed -- but if some university/group is willing to support this, it can really improve things. We could additionally also choose for what kind of packets (data) to enable this for.
Multipath-tcp could potentially be useful on the LAN side as well, if there is IIAB and multiple routers involved. It could smartly route traffic between user-user and user-server. However, at this point, I am not very sure if it offers any tangible improvements over layer 2/3 meshing protocols such as batman-adv et. al.
The text was updated successfully, but these errors were encountered: