[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: relationship of snmpconf to midcom?

What I'm saying now is my understanding of midcom which may or may not be
understanding of a prospective WG, IESG, or whomever else.

The driver behind midcom is the desire to get VoIP and other applications
using session protocols (SIP, RTSP, H.323, ... ) across firewalls
and NATs (and possibly other "middleboxes"). It is difficult unless
the middleboxes understand the applications. Though one may have 
embedded ALGs in such middleboxes, there are scenarios in which
it seems easier to have the application awareness out of them.

This is the scenario midcom is looking at. A protocol is needed 
which allows the application to ask "please, tell me a NAT translation 
that will be applied to my packet voice stream" or "please, open a pinhole 
for the stream in the firewall".

There are, IMHO, two items on the midcom to-do-list: to document this
problem: scope, issues, requirements, etc. Once the problem is understood 
and  documented, the next item will be to show how it can or cannot be 
solved using existing protocols. Meanwhile, I've heard many protocols 
spelled in this context including SNMP, COPS-PR, LDAP, MGCP, .... 

To address some of the technical comments/questions raised previously:
- It will be a requirement that the midcom protocol (whatever it is)
  operates in real-time.
- I do not think that web caches are of primary interest. Firewalls and 
  NATs are.

Those interested in foglamps may check its archive and/or subscribe to
the mailing list:

To subscribe, send email to majordomo@lists.panix.com with “subscribe foglamps” in the body of message
Archive: http://www.fokus.gmd.de/glone/ietf/foglamps/


"Joel M. Halpern" wrote:
> I should be rather surprised if there was any relationship at all.  From
> reading Elliot's notes, Midcom is about defining architecture (and maybe
> even protocols) which get in the middle of the users data streams.  The
> most obvious example is web caches and proxies.  I believe that they also
> include NATs and Applications gateways in the problem space.  None of this
> has anything to do with SNMP or SNMP for Configuration.
> Yours,
> Joel M. Halpern
> At 04:15 PM 1/5/01 -0800, Randy Presuhn wrote:
> >Hi -
> >
> >Does anyone know how this relates to the snmpconf work?