Re: Some definitions (Was: Re: New issue 16: No packets fromother end?)
"James Kempf" <[email protected]>
| Newsgroups | gmane.ietf.mobike |
|---|---|
| Message-ID | <[email protected]> |
> > Actually, SCTP doesn't send packets over multiple paths either. It uses
the
> > addresses only for failover, that is when it detects a failure, it
switches
> > to another address.
>
> Yes, I think you are right about the usage of just one path at a
> time. Question: does SCTP still have
> some specific congestion avoidance for multi-homing, such as
> going back to slow start upon changing to a different path?
>
Depends on the amount of time the particular address pair is idle. This from
Section 7.2.1 of RFC 2960:
Beginning data transmission into a network with unknown conditions or
after a sufficiently long idle period requires SCTP to probe the
network to determine the available capacity. The slow start
algorithm is used for this purpose at the beginning of a transfer, or
after repairing loss detected by the retransmission timer.
o The initial cwnd before DATA transmission or after a sufficiently
long idle period MUST be <= 2*MTU.
jak