Hi folks,
I would just like to add that our studies are ongoing, and
this:
variety of Internet paths. It observes that on average, IP options
introduce between 7% and 26% (for different sets of paths at different
times) of additional delay. The article says that "In any case, the
additional delay is clearly an order of magnitude smaller than the
RTT."
is only a part of our findings, which are only a part of
all the things that *should* be looked at. As Simon suggests,
processing overhead in routers may be an issue if they are
flooded with packets carrying options. Moreover, we also
saw that this:
3.) Operators configure their routers to ignore IP Options.
and this
4.) Operators configure their routers to drop packets with IP Options.
is already happening quite a bit nowadays.
Our findings are also limited to NOP options vs. no option
at all, and ICMP packets, not TCP. For different (even
more disappointing) findings with TCP and a new option
number, see pages 9/10 of:
http://www.icir.org/mallman/papers/tcp-evo-submit.ps
... who knows what's happening to packets that carry RA.
I believe that options are a very reasonable (and certainly
the most flexible) means of communication between end systems
and routers by design, and they could theoretically be used
for lots of great things in the future. The point of my
ongoing measurement effort is to show that they might not
be absolutely useless in the current Internet, and give
folks some concrete numbers - I don't believe that options
are ALWAYS useless (but don't ask me for a
counterexample :( ); depending on the envisioned usage
scenario, the measured properties will make things look
good or bad.
Cheers,
Michael
_______________________________________________
Ietf mailing list
Ietf(_at_)ietf(_dot_)org
https://www1.ietf.org/mailman/listinfo/ietf