ietf
[Top] [All Lists]

Re: Any comparison Study on MGCP vs H.323, MGCP vs SIP

2000-05-15 11:30:02
Sez "Masataka Ohta" 
<mohta(_at_)necom830(_dot_)hpcl(_dot_)titech(_dot_)ac(_dot_)jp>
H.323 is defined for a LAN environment, not for telephone lines.

For telephony people, the IP protocol is for a LAN environment
that there is no difference between H.323, SIP, TELNET, or DNS
for that matter.

"Telephony people" are not relevant here, since we're talking about
VoIP.

As I said:

For VoIP over telephony networks

Your statements don't make sense; "VoIP over telephony networks" is an
oxymoron, since VoIP is, by definition, over an IP network.

some protocol is necessary between two telephony networks.

FYI, there is a protocol called TBGP proposed in IETF for the purpose.

TBGP is the proposed method for binding E.164 numbers to telephony
domains, much like DNS binds names to IP addresses.  One would still use
SIP (or equivalent) to actually complete the call.

If you want to use ITU protocols, please choose some other numbers.

So, you are saying SGCP/MGCP are wrong to use 323.

Fine.

No, he's suggesting that if you wish to use an ITU protocol, H.323 is
not the correct one.  Perhaps Q.931?

FYI, in my design of "The Simple Internet Phone":

If you are interested in Internet telephony, see you at
INET'2000 in Yokohama for the presentation of our paper
"The Simple Internet Phone".

Please let us know the URL where you'll be publishing this paper, as
some of us may not be inclined to fly to Yokohama to hear about yet
another non-standard, proprietary telephony protocol.

I chose to keep using 164 (sorry, not 42) at least for the time being,
because it is the easiest way to let subscribers replace telephony
network with the Internet.

MGCP/SGCP/Megaco directly use E.164 numbers.  SIP allows users to see
only E.164 numbers during a transition period, though it becomes much
more pwoerful when you move to a more expressive namespace.

[from a prior message]
As I pointed it out with regard to iMODE and WAP, an attempt to
promote protocols like SIP, a NAT friendly protocol even more
complex than H.323

SIP may or may not be NAT-friendly, a point which is best left to other
(time-wasting) threads.  I would love to see any explanation of how SIP
is more complex than H.323; maybe you have them backwards?

was based on a wrong strategy destroying
the Internet into a collection of mostly-non-IP networks connected
by application/transport gateways with mostly-non-IETF
application/transport protocols.

SIP is not, as you state, based on a strategy of building non-IP
networks and connecting them with non-IETF protocols; in fact, it's
quite the opposite.  SIP allows the replacement of non-IP (ie. legacy
telephony) networks and non-IETF (ie. ITU) protocols; in the ideal SIP
world, legacy telephony would cease to exist.

While a bit dated, Henning Schulzrinne and Jonathan Rosenberg's paper
has quite a bit of detail on the subject:
http://www.cs.columbia.edu/~hgs/papers/Schu9807_Comparison.ps.gz

Masataka Ohta

S

     |          |         Stephen Sprunk, K5SSS, CCIE #3723
    :|:        :|:        Network Consulting Engineer, NSA
   :|||:      :|||:       14875 Landmark Blvd #400; Dallas, TX
.:|||||||:..:|||||||:.    Email: ssprunk(_at_)cisco(_dot_)com