On Tue, Aug 12, 2008 at 10:17:32AM -0400, Henning Schulzrinne wrote:
>> - In congested networks
>
> The SCTP congestion control algorithm is essentially the same as TCP, so 
> there's no advantage. Unfortunately, there seems to be a fair amount of 
> mystique around SCTP which isn't borne out by the observable facts.

There might be subtle differences, e.g., whether one counts acknowledged
packets or bytes, and there may be implementation specific "optimizations",
which cause observable performance differences.

Some time ago I measured signaling message transmission delays over an
uncongested link with synthetic packet losses. For medium to high
message rates, Linux SCTP with one stream and ordered(!) transmission
(i.e., TCP-like transport) caused significantly smaller delays than
Linux TCP, and was closer to the values calculated using a model of 
fast retransmit and head-of-line blocking. For lower message rates
TCP performed better. Seeking for explanations one has to deal with
issues such as usage of ABC, or deriving RTO from RTT estimates, etc.



Coming back to the original question, I also doubt that one would see
advantages in the real world, which justify the additonal effort.

regards,
Sebastian

-- 
Sebastian Kiesel            mailto:[EMAIL PROTECTED]
Network Research Division   tel:+49-6221-4342-232   fax:+49-6221-4342-155
NEC Laboratories Europe     Kurfuerstenanlage 36, 69115 Heidelberg, Germany
--
NEC Europe Limited          Registered in England 2832014
Registered Office           NEC House, 1 Victoria Road, London W3 6BL
_______________________________________________
P2PSIP mailing list
[email protected]
https://www.ietf.org/mailman/listinfo/p2psip

Reply via email to