RE: Loops (RE: CPIM changes)

I do understand that your argument is that CPIM can work without loop
detection, and that based on your experience with email, we can live without
requiring a standard attribute/mechanism. However, from my perspective the
risks are plain and they appear impactful. I think the respondents to this
thread so far have all suggested that IM would not 'work' without loop
detection - it is more than 'helpful' to be able to prevent a plausible
misconfiguration in endpoints from bringing down your gateways. Regardless
of whether or not email had standard loop detection for all of its lifetime,
it does have it today - which suggests to me that we are ignoring its lesson
(turning back the clock on our understanding by thirty years) if we neglect
to address looping. 

Are there any gateway practices other than mapping some sort of hop-count
indicator between protocols that we could recommend to reduce this risk? It
does seem possible that a gateway could recognize when a request has looped
back to itself (I send a request out to a SIMPLE network and receive the
same request back). However, if multiple gateways (for scalability purposes)
are used between peering IM networks (allowing a looped request to arrive at
a different gateway), or triangular forwarding of some kind occurs, this
wouldn't help. But perhaps it's better than nothing.

Jon Peterson
NeuStar, Inc.

> -----Original Message-----
> From: Dave Crocker [mailto:[email protected]]
> Sent: Thursday, November 14, 2002 1:13 AM
> To: Peterson, Jon
> Cc: 'Dave Crocker'; Mark Day; [email protected]; 
> [email protected]
> Subject: Re: Loops (RE: CPIM changes)
> 
> 
> Jon,
> 
> Thursday, November 14, 2002, 12:39:43 AM, you wrote:
> Jon> ... but I think we already dictate quite a bit (all of 
> the attributes and
> Jon> operations in impp-im-00), and in fact we -must- 
> guarantee a certain level
> Jon> of commonality if any gatewaying is to be possible - 
> this is really a
> Jon> question of whether or not we consider loop detection to 
> be a 'core'
> Jon> function that we must dictate.
> 
> can it work without it?  the answer is yes.  so it's not core.
> 
> is loop detection extremely helpful?  the answer is also yes. 
>  but helpful
> is not the same as core.
> 
> d/
> -- 
>  Dave Crocker  <mailto:[email protected]>
>  TribalWise <http://www.tribalwise.com>
>  t +1.408.246.8253; f +1.408.850.1850
> 



  [reminder: [email protected] for non-technical discussions, please]