Re: [quicwg/base-drafts] Idle timeout interaction with RTO (#1429)

[ianswett <[e-mail]>]

Chromium now has them separated(see: the PathDegradingAlarm and the RetransmittableOnWireAlarm https://cs.chromium.org/chromium/src/net/third_party/quic/core/quic_connection.cc?sq=package:chromium&g=0&l=281)  Path degrading is there solely for connection migration at the moment, and RetransmittableOnWireAlarm really should be able to be combined with the existing ping alarm.

You're right that separating idle timeout from dead path detection is a good choice in the long term.  We might only want to wait 15 seconds before declaring the path dead when we're retransmitting data, but a connection may have a 5 or 10 minute idle timeout.

My concern about dead path detection is that I don't have enough data to support a particular choice of when to consider a path dead.  5RTOs worked fairly well in the one experiment, but I have no comparison data with 4 or 6RTOs, and all of these are magic numbers.

-- 
You are receiving this because you are subscribed to this thread.
Reply to this email directly or view it on GitHub:
https://github.com/quicwg/base-drafts/issues/1429#issuecomment-396431948