begin  quoting Andrew Lentvorski as of Mon, Apr 17, 2006 at 08:57:31PM -0700:
> Stewart Stremler wrote:
> >I'm trying to say (and not well, it seems) that any pain suffered by
> >making NAT work with a reasonable protocol is _exactly_ _the_ _same_ _pain_
> >as making that protocol work with a default-deny protocol: you have to
> >go poke the device to say "let connections through to _this_ machine at
> >_this_ port".  Consequently, NAT is just one extra step, or, if you
> >have your firewall assume the responsibility, it's not even that.
> 
> Not everyone wants "default deny all outbound" on their network.

True.  But if we're designing systems that won't work with such a
policy on the firewalls, that sort of policy is essentially *dead*.
The choice is taken away from me, while protocols that *do* work with
such firewalls don't impose that sort of constraint on anyone else.

> That's a religious issue.  We are going to have to agree to disagree.
 
Heh.

Fair 'nuff.

> >And UDP packets are dropped before TCP packets anyway. (I suppose that
> >might result in lots more retries and resends, because acknowledgements
> >won't get through... but any router that has a congestion problem has
> >an easy solution: drop more UDP packets. No more congestion.)
> 
> Can I get a reference?  I haven't heard of routers dropping UDP in 
> preference to TCP.  That would require packet inspection.  Now, UDP 
> won't retry, but that doesn't mean that the routers choose TCP over UDP.

I've been looking, and I can find neither the notes nor the reference
book.  You're probably right, and I'm misremembering, and it was an
effect, not a decision.  If I can find my notes on the topic, I'll
follow up.

(Too. Much. Junk.)

[snip]
> Now, I can certainly question the validity of those numbers, but it 
> seems to be in the right ballpark.  The issue is that torrents move a 
> *lot* of data.  The fact that ISPs are buying expensive traffic shaping 
> boxes indicates that this is indeed large enough to notice.

Hm... I see.

Of course, Tracy often has some interesting comments on torrents... they
aren't as useful for him as one would expect.
 
> >TCP is hard to replace ... you end up implementing something that looks
> >an awful lot like TCP.
> 
> Absolutely.  It looks like TCP but goes through NATs.  Good enough.

UDP doesn't go through NATs any better than TCP does. Same issues apply.
I must be missing your point here.

> And we all know what happens when something is "good enough".

Heh. Yup.

-- 
_ |\_
 \|


-- 
[email protected]
http://www.kernel-panic.org/cgi-bin/mailman/listinfo/kplug-list

Reply via email to