[H.323-SIP/Internet] The reason.

Paul E. Jones paul.jones at TIES.ITU.INT
Sun Feb 27 13:40:50 EST 2000


Francois,

I do not take your arguments personally.  But my statements are statements
of fact, nothing more.

Allow me to comment on your statements.

> Paul wrote:
> > It's not an impossible task, but one that may lead to tremendous debate.
It
> > is quite obvious that some members of the ITU and some members of the
IETF
> > have very basic philosophical differences.  I can tell you that some
members
> > of the IETF will quickly reject anything the ITU does to standardize
> > interoperability.  I can also tell you that some of those members will
also
> > reject anything ETSI does, as well.
> >
>
> Couldn't you think that this may actually have something to do with the
> fact that 3-4 years of interoperability trials for H.323 in Etsi/ITU/IMTC
> have lead to an amount of interoperability that was acheived in 1 or 2
> bake-offs for SIP.

First of all, I should probably make it very clear that I have worked on
both H.323 and SIP projects.  I am aware of the stengths and weaknesses of
each.  There is a generally feeling that SIP is so easy and H.323 is much
harder.  In some ways, that's true-- in particular, the ASN.1 PER requires
work.  However OSS provides an excellent set of tools for that, so that's
really a non-issue.

H.323 supports multipoint, multimedia conferences.  With H.323, there is the
concept of a "conference" in which voice, video, and audio may be used.
SIP, on the other hand, is a "session initiation protocol".  It is not a
conferencing protocol.  In fact, since the BYE message is optional, SIP
doesn't even exist after the remote party accepts the INVITE message and the
ACK is returned.  From that point forward, the "session" is nothing more
than RTP streams flowing.

There have been many H.323 interoperability events.  Yes, there have been
more problems than at SIP bake-offs, but that's because H.323 is a more
complete system with a little more complexity to support multipoint voice,
video, and data.  But those interop events continue, because the standard
continues to grow and evolve-- not because all of the previous events were
total failures.

Presently, there are literally millions of H.323 devices out there in the
world in production networks.  There are major carriers using H.323 to
provide VoIP-- it's real.  As far as I know, there is not one single
production network running on SIP.  Perhaps Level 3 might be using it
between MGCs, but that deserves no special bragging rights.  Perhaps I am
wrong, but whatever SIP networks exist pales in the shadow of H.323.

> That going to a SIP backeoff is cheap and attending
> the ITU-T is outrageously expensive.  You can try to depict otherwize, but
> the basics underlying reasons why the ITU-T is doomed to perpetuously
> deceive IETF participants are still there ... no rough concencus, no
> running code, membership fees which lock-out small players, closed-policy
> on publication of standards, closed mailing lists, shall I go on ?

The ITU does require concensus and that often leads to a larger protocol.
Essentially, some folks want one set of features and another group wants
another set.  The solution-- if they don't conflict, we add both.  However,
those additions are almost always optional.  Essentially, manufacturers get
to introduce into the standard what their customers want or need.

As for the price of the ITU... it is a bit expensive, but we do not exclude
the public from all meetings.  The ITU meetings in Geneva are only for
sector members or represetatives of a delegation.  If any company wanted to
attend the ITU meeting and they were not a sector member, the could most
likely contact their government agency (U.S. State department in the US) and
ask permission to attend.  Most likely, permission would be granted.

During the year, we also have "Rapporteur meetings".  Those are completely
open to the public and we have had IETFers at those meetings.  The mailing
list is certainly not closed: Intel runs the mailing list!

As for the cost of the ITU: I have no control over that.  The ITU, being
part of the United Nations, has a responsibility to make documents available
in multiple languages and provide meetings with interpreters.  There are
costs associated with that.  I suppose the fees charges are necessary in
order to maintain their facilities and employee the personel necessary in
order to produce the various translations.

> The interest in interoperability with SIP is presently a  one way
> relationship which comes from the H.323 people. Thus, it is the same
> H.323 people whi should consider coming to an industry-open,
> SIP bake-off.  You can try talking Henning Schulzrinne into lettin you
> in. I'm sorry, but we just dont need the ITU-T, ETSI nor the IMTC slowing
> us down at the IETF ;-)

Actually, the first H.323/SIP interworking effort was started on the IETF
side.  I'm not sure who started the list, but I know that Henning publicized
the mailing list sip-h323 at eGroups.com.  I believe it's his list.

Your last comment is precisely the kind of attitude I was talking about when
I said that "I can also tell you that some of those members will also reject
anything ETSI does, as well."  It's interesting that members of the ITU will
adopt specifications from the IETF or other organizations-- whatever is
necessary to meet the requirements-- but that so many IETF people maintain a
"holier-than-thou" attitude and are so unwilling to work with other
standards bodies.

> I think that all of your analysis is rubbish.  You could simply bring your
> stuff at one of the SIP bake-offs and get the feedback you need to go on
> and experiment a little more with H.323 ... You can try all you want to
> turn a Gatekeeper into a Proxy server/Redirect server, but it'll probably
> not happen in the marketplace.

I believe that you have reconfirmed my analysis.  Honestly, though, it's not
"analysis".  It's experience.  I have attended IETF meeting and would
continue if I had the time.  I work with people who do attend the meetings.
I know a lot of folks within the IETF and the ITU and I can tell you that
there are very basic philosophical differences between the two
organizations.

As for turning a GK into a proxy server, it's more the other way around: a
proxy server is looking more and more like a Gatekeeper these days.  And
will you see them converge?  Quite possibly-- certainly, they are converging
functionally.

> This said, I'm sorry for flaming you.

I take no offense.  But I do believe this exchange (and the recent
H.248/MEGACO fiasco) and demonstrated that it is very difficult for the two
organizations to work cooperatively.  I am still open to a cooperative
effort, because I think that is the best thing for our customers.

Best Regards,
Paul E. Jones
Editor, H.323
Cisco Systems, Inc.



More information about the sg16-avd mailing list