[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: Comments on draft-ietf-shim6-failure-detection



Bernard Aboba wrote:

The "green light" is carrier sense. While loss of carrier sense can be handled immediately on wired networks, it would not be wise to treat it the same way on a wireless network.
Yes, some damping process is needed.

(But I believe the action depends also a little bit on the kind
of network we are talking about, and what specific green
light we are referring to. E.g. you know that your access
auth state is not where what it should be, you can't get
a PDP context or the network told you to go away vs. just
not hearing the other end.)

But for the rest, we can only do operations that are relatively
slow. My view of what's achievable is somewhere between
RTO and the time TCP gives up. Interestingly, on this timescale
TCP has probably already slowed down.

Right. In RFC 3539, this is characterized as somewhere between two and four retransmissions. Assuming RTO = RTOmin (1 second), this yields a timer between 7 seconds (1 + 2 + 4) and 31 seconds (1 + 2 + 4 + 8 + 16).
Right.

However, this assumes that the protocol is not running over wireless WAN links such as GPRS. In such a situation, 7 seconds might only be time for a single retransmission.
I'm getting sidetracked from the initial issue, but it may
be interesting to note that cellular network RTTs have
been decreasing and continue decrease substantially
from the initial GPRS deployments. If I remember correctly
modern network type and proper equipment gets you
in the 100ms range. YMMV though.

--Jari