[standards-jig] Pub/Sub for JNG?

Joe Hildebrand jhildebrand at jabber.com
Wed May 1 21:01:08 UTC 2002


OK, now it's my turn to take a crack at the troll.  At this
point I'm assuming that Dave is having fun getting us all
riled up.


Congestion control isn't about making your UDP connection
work faster.  It's about letting everyone else have a crack
at the bandwidth available.

>From http://www.psc.edu/networking/tcp_friendly.html 
(the I'm Feeling Lucky choice from
http://www.google.com/search?q=congestion+control+tcp):

<blockquote> 
- Applications which perform congestion control make more
efficient use of the network and should generally see better
performance because of it.

- Applications which adapt to the network are capable of
running over a much wider range bandwidths and are hence
more useful in the Internet.

- Congestion control algorithms prevent the network from
entering Congestive Collapse. Congestive Collapse is a
situation where, although the network links are being
heavily utilized, very little useful work is being
done. (Think of metropolitan traffic gridlock...)

- The network will soon begin to require applications to
perform congestion control, and those applications which do
not perform congestion control will be harshly penalized by
the network (probably in the form of preferentially dropping
their packets during times of congestion).
</blockquote>

The last point is very important.  
The IETF Transport Area Working Group
(http://www.ietf.org/html.charters/tsvwg-charter.html) is
coming down **hard** on any protocol that does not specify
congestion control.

I assert that once you add in adequate congestion control
and real, honest-to-gosh reliability, not to mention a
little bit of security, your UDP protocol will be bulkier
and slower than TCP.  We've been working on TCP for 20+
years, and it's been getting smaller and faster all the
time.  There's not much in there that isn't needed.

I'm willing to believe that there are protocol geeks (term
of respect) out there who know more about queuing theory
than I do.

On another note, TCP is a lot easier to get through
firewalls, because of all of that "overhead".

Now, if you wanted to recommend SCTP
(http://www.ietf.org/rfc/rfc2960.txt), you might have a good
argument, as long as you first get all of the OS, firewall,
and router vendors to support it.  :)


Dave <dave at dave.tj> writes:

> If TCP were so quick and efficient, I wouldn't be able to write a simple
> (unoptimized) FTP implementation (client+server) using UDP that can outrun
> my UNIX command-line FTP connecting to my ftpd by 10%-30%.  I'm sure
> a good programmer can implement FTP over UDP even better than I did.
> If speed is your drug, TCP is the equivalent of a cop.
>
>  - Dave
>
>
> Peter Millard wrote:
>> 
>> Dave -
>> 
>> It's apparent that you have no idea about the goals and extent of TCP ...
>> I'd suggest you read the following links before you post any more replies:
>> 
>> http://www.ietf.org/rfc/rfc793.txt
>> http://www.ietf.org/rfc/rfc768.txt
>> http://www.ietf.org/rfc/rfc791.txt
>> 
>> pgm.
>> 
>> _______________________________________________
>> Standards-JIG mailing list
>> Standards-JIG at jabber.org
>> http://mailman.jabber.org/listinfo/standards-jig
>> 
>
> _______________________________________________
> Standards-JIG mailing list
> Standards-JIG at jabber.org
> http://mailman.jabber.org/listinfo/standards-jig

-- 
Joe Hildebrand
Chief Architect
Jabber, Inc.
http://www.jabber.com/




More information about the Standards mailing list