On Wednesday 02 Mar 2011 18:22:52 xor wrote: > As load management is something which has been scientifically investigated > for > decades I vote for sticking to established standards first and only start > homebrewing if they prove to yield very bad results. > "Very bad" should be measured by comparing to their expected results in > theory. There are nice graphs out there and nice books about TCP etc. > - Nobody would complain if you spent some FPI money on getting a popular book > on load management. > You are confusing load management with congestion control.
The recent changes were straight out of the TCP-related RFCs, although arguably they don't apply directly as TCP is in-order and therefore acks are cumulative. However TCP allows very high latency in such cases. Whereas we have some fairly short timeouts at the layer above. The two do not mesh cleanly!
signature.asc
Description: This is a digitally signed message part.
_______________________________________________ Devl mailing list [email protected] http://freenetproject.org/cgi-bin/mailman/listinfo/devl
