- From: Jeffrey Mogul <mogul@pa.dec.com>
- Date: Fri, 26 Apr 96 16:00:17 MDT
- To: Anselm Baird-Smith <abaird@w3.org>
- Cc: http-wg%cuckoo.hpl.hp.com@hplb.hpl.hp.com
I am having a little problem with the proposed computation for the
correction to the received_age. Draft02, p. 87 states:
corrected_initial_age = corrected_received_age + (now - request_time)
If I don't miss something: 'now-request_time' measures the round time
trip between the receiver and the sender, while what we want to
measure is only the time between the sender and the receiver (since
the Age value at the sender is supposed to be updated right before the
sender emits it - see paragraph 5 of the same section).
While I understand that as an approximation, this is on the safe side,
I would like to know if this was intentional (in which case it 's
probably safer to state it in the text).
This *is* intentional. The Age calculations are meant to be as
conservative as possible, and because it is not true that all
Internet paths are symmetrical, the most conservative assumption
is that all of the round-trip delay is in the server-to-client
direction.
There are existing examples of highly assymetrical paths in the
Internet.
-Jeff
Received on Friday, 26 April 1996 16:11:14 UTC