Re: [tcpm] Variable "recover" in F-RTO
[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [tcpm] Variable "recover" in F-RTO



Hi Alex,

Sorry that it took this long to reply.

The condition 'but not more than "recover"' is there just to cover the case of invalid ACKs arriving for some strange reason, and in such case revert to normal TCP operation, to be on a safe side. In a well-behaving environment one should not receive such an ACK.

- Pasi


On Mar 5, 2009, at 2:37, ext Alexander Zimmermann wrote:

Dear all,

I have an understanding problem with aforementioned variable "recover"
in the current version (http://www.tools.ietf.org/html/draft-ietf- tcpm-rfc4138bis-04)
of the F-RTO Algorithm.

Section 2.1 The Algorithm (F-RTO, Reno/NewReno Case)

Step 2): When the first ACK after RTO rexmit arrives at the TCP Sender, we store the highest sequence number transmitted so far in the variable
"recover".

then

Step 2a) Check if the ACK from Step 2) is a DUPACK OR the ACK covers
"recover" but not more than "recover"...


Why is the check like this? IMHO we can only get an ACK that cover exactly "recover" (Lost Retransmission - SIGCOM Paper Section 4.2) or an ACK that cover less than
"recover", but we cannot get ACK that cover more than "recover".

Alex


//
// Dipl.-Inform. Alexander Zimmermann
// Department of Computer Science, Informatik 4
// RWTH Aachen University
// Ahornstr. 55, 52056 Aachen, Germany
// phone: (49-241) 80-21422, fax: (49-241) 80-22220
// email: zimmermann at cs.rwth-aachen.de
// web: http://www.umic-mesh.net
//



Note: Messages sent to this list are the opinions of the senders and do not imply endorsement by the IETF.