Re: TCP end-to-end Semantics

From: Mark Allman (mallman@grc.nasa.gov)
Date: Tue Jan 09 2001 - 13:01:46 EST

  • Next message: Anil Agarwal: "Re: TCP end-to-end Semantics"

    > If you linger on the close (such that all data is acknowledged
    > before your socket closes -- and assuming no ack spoofing) then
    > you'll have a firm guarantee that your data reached the remote
    > system.

    Modulo LINGER working correctly (which my experience suggests is an
    iffy proposition).

    > This guarantee, however, is only for the end system -- it does not
    > say the end application received the data (it may never read it).

    I believe the strict application of the end-to-end principle is the
    only true answer. That is, the applications have to ACK the arrival
    of the data to its peer application. Trusting anything (transport
    protocols, routers, spoofing boxes, proxies, NATs, etc.) in between
    the application processes for "guarantees" is always going to be
    dicey (where the defintion of "dicey" sort of depends on what is
    involved in between the apps).

    allman

    ---
    Mark Allman -- NASA GRC/BBN -- http://roland.grc.nasa.gov/~mallman/
    



    This archive was generated by hypermail 2b29 : Tue Jan 09 2001 - 13:38:37 EST