On Tuesday 16 Apr 2013 16:38:28 Vladislav Sterzhanov wrote: > Hi there, FreeNet people. > I am a student from a Belarusian State University and I'd like to > participate in this year's Summer of Code by improving the Free Network. > Generally, I wish to enhance the survivability of slow and dying > connections, while boosting the broad ones. > Some key points that I see after viewing Ideas page and searching the bug > tracker: > - providing the ability for dynamic detection of the currently available > MTU size along the path to the node, making it possible to forecast nearly > exact suitable packet size for the connection. (issues > 0000976<https://bugs.freenetproject.org/view.php?id=976> > , 0005294 <https://bugs.freenetproject.org/view.php?id=5294>)
Sounds good. Any ideas how to implement this? > - dealing with the low-bandwidth and high-latency links by more precise > control over them - splitting packets to get them through, estimating the > appropriate wait-timeout, etc. > (0002123<https://bugs.freenetproject.org/view.php?id=2123> and, > partly 0005630 <https://bugs.freenetproject.org/view.php?id=5630>) > This two could be significantly improved by tracking the connection stats > (sent packets / retransmits / acks) for the nodes that we're currently > connected to. Provided with such information and making some empirical > assumptions we could additionally adjust estimations needed and determine > the reliability of the connection. Right. We're vaguely TCP-like but the current congestion control performs rather poorly. > - auto-adjusting the acknowledge method for a connection based on > available information (priority for poor connection is to send and receive > at least something, whereas for low-latency one additional savings provided > by cumulative zero-ack would be a good idea). > - some more high-level control also seems feasible: e.g. mentioned > idle-priority chunks can reduce the additional load on network at > rush-hours. That would be nice, we don't have anything to use it yet, but I suspect we could put together something simple fairly quickly. > > Perhaps I got something wrong, moreover, I bet that there is much more work > to do that I've already found out, so I need to know what do you think > about it? Yes, if you have some experience with networking code, we'd be very happy to have you as a SoC student looking at these issues. Anyone want to mentor Vladislav/Quadrocube? However do bear in mind that we'll need to see you fix a bug or a small feature *before* we accept your application. Thanks!
signature.asc
Description: This is a digitally signed message part.
_______________________________________________ Devl mailing list [email protected] https://emu.freenetproject.org/cgi-bin/mailman/listinfo/devl
