[ih] IEN Notes and INWG

[John Day]

We have a paper in the works in which we have looked at a number of 
attacks on protocols of this type and delta-t is not susceptible to 
them.  Basically it comes down to delta-t decouples port allocation 
from synchronization.  TCP combines the two and then overloads port 
allocation with application naming.

I was quite surprised, because like you, I assumed that since delta-t 
had not been subjected to that kind of hardening, it would have the 
same vulnerabilities.  Turns out not to be the case at least when 
used in the context we are looking at it.

We haven't looked at all of them yet, but so far the results are 
quite surprising.

We are looking at it in an architecture does not do listens on 
well-known ports.  A short term hack that should have been gotten rid 
of decades ago.

Take care,
John

At 18:51 +0000 2010/03/23, Tony Finch wrote:
>On Mon, 22 Mar 2010, John Day wrote:
>>
>>  TS was simpler than TCP, but still not the answer.  That honor goes to
>>  delta-t, which is not only simpler but also has better security properties
>>  than both.  (Watson's proof that bounding 3 timers is necessary 
>>and sufficient
>>  is one of the most important results (as well as surprising and 
>>astounding) in
>>  all of networking and little understood or recognized. An incredible
>>  intellectual achievement.)
>
>As far as I can tell from the literature, delta-t had no hardening against
>denial of service attacks. The server must be prepared to buffer data from
>the client before it can prove that the client is listening. So I'm not
>sure in what sence it is more secure.
>
>Tony.
>--
>f.anthony.n.finch  <dot at dotat.at>  http://dotat.at/
>GERMAN BIGHT HUMBER: SOUTHWEST 5 TO 7. MODERATE OR ROUGH. SQUALLY SHOWERS.
>MODERATE OR GOOD.