Re: [Sipping] P2P why not DNS for discovery?

[Cliff McCollum <cliff at voicemobility.com>]

My concern with using DNS is pretty simple: it doesn't always have the entries I'm looking for. Two simple examples:

1) I have a small office with a half-dozen machines. We don't run our own DNS, because we have no need for it, and the individual machines aren't in the upstream DNS because we're behind a NAT firewall. However, I want to use P2P SIP in order to communicate from office to office. If I am dependent upon DNS to house my URI's, I'm out of luck.

2) I'm running a bunch of gizmos all connected via BlueTooth or WiFi (my cell phone, my PDA, my Gameboy :-) These gizmos won't end up the DNS anytime soon (maybe in the world of IPV6?), but they should still be able to find each other using P2P SIP.

Cliff



Brian Rosen wrote:

I'm not sure I buy this argument.
DNS is sufficiently distributed that I wouldn't consider it centralized, and
its real life situation is that it is highly reliable and highly available,
and of course, essentially free (you're paying for DNS service somewhere,
but it's usually built into your access bill). 

If the solution relied in dynamic DNS, then we have a little less of the
"highly reliable" and "highly available" and "free".  You can get one of
those for sure, and two out of three often, but usually not all three.
Nevertheless I suspect even dynamic DNS is good enough.

I'm not much of a purist.  I think Juha's ideas might work and merit more
thought.

Brian



  

-----Original Message-----
From: sipping-bounces at ietf.org [mailto:sipping-bounces at ietf.org] On Behalf
Of Brijesh Kumar
Sent: Thursday, February 03, 2005 9:10 AM
To: Juha Heinanen
Cc: sipping at ietf.org
Subject: Re: [Sipping] P2P why not DNS for discovery?

Juha:

----- Original Message -----
From: Juha Heinanen <jh at tutpro.com>
Date: Wednesday, February 2, 2005 9:21 pm
Subject: [Sipping] P2P why not DNS for discovery?

    

there must be some ery good reasons for inventing all those new p2p
protocols we have read on this mailing list, but i would still
like to
hear why an existing protocol, namely DNS, is not good for discovery?
      

DNS is good for discovery, but is it P2P?

P2P applications need to  run without relying on centralized servers in
networking
infrastructure. So, clearly the use of centralized DNS to resolve does't
fullfill
that requirement. Actually, there is a project called Distributed DNS
which is
a P2P version of DNS based on DHT. While P2P based DNS is useful idea, it
is
is n't designed or meant to replace sip registrar function. The efficiency
of
DHT based search, makes P2P sip a good proposition to explore.

    

my sip uri is sip:jh at tutpro.com.  if someone wants to get my contact
information, they could simply make a sip srv query on jh.tutpro.com,
which my sip ua is keeping up to date via dynamic dns updates either
directly or indirectly using xmlrpc, for example.

what is wrong with this kind of DNS approach?
      

Nothing wrong. But, this clearly is not P2P approach being
pursued in P2P SIP.

cheers,

--bk

    

-- 
Cliff McCollum, M.Sc.
Vice-President Technology and Development, Voice Mobility Inc.

cliff at voicemobility.com
www.voicemobility.com
Voice, Mobile, Fax: (250) 978 4255

_______________________________________________
Sipping mailing list  https://www1.ietf.org/mailman/listinfo/sipping
This list is for NEW development of the application of SIP
Use sip-implementors at cs.columbia.edu for questions on current sip
Use sip at ietf.org for new developments of core SIP