Err, I think there are some things missing:
1) H.323 closely matches PSTN protocols and capabilities. Its
interoperability with ISDN and SS7 are far more natural.
2) H.323 is more efficiently coded using ASN.1. One might not think that
this matters, but in fact it matters a great deal in large volumes.
H.323 is PER encoded, which frequently results in about 50% compression
over uncompressed data. Smaller packets have lower latency, and take less
time to transmit. Advances in processor speed over IO speed has made
compression almost a zero cost feature.
3) Standardization is _way_ better with H.323. The IETF process relies on
the standard-writers to do the implementations, which we then hope are
well documented in the standard. They aren't always so well documented. So
long as the group of implementers is small, this isn't a problem. But in
large groups, it doesn't work very well. Such an iterative process of
implementation/documentation/implementation is slower than complete
specification -> implementation except when the group of implementers and
documenters are small, so that undocumented or vague parts can be passed
informally amoung the implementers. For example, consider that the DNS
groups have been struggling for years with ambiguities in RFC 1035.
Things still aren't fully specified, and there is a great deal of
difference and incompatibility between Bind 9 implementation and other DNS
implementations. It doesn't seem that these ambiguities will be resolved
4) H.323 has free infrastructure: There are free ASN.1 compilers (I'm even
am about to release one). There are free implementation of H.323 protocol
stacks, and free h.323 software.
5) H.323 moved through 4 major well defined enhancements far faster than
6) SIP started out saying that it could do the job simpler and easier than
H.323. So far, this hasn't proven to be the case. SIP is now just as
complicated as H.323. The reason for the complication is driven mostly by
the PSTN, and the features that you find on phones and PBX's. These
features haven't changed, so the complexity is still the same.
However, this is not to say the SIP effort was wasted. Not in the least.
There has been a tremendous amount of productive feedback and cross
pollenation between the two groups. There is nothing like competiton to
make one think pragmatically. SIP significantly influenced the attention
to performance in H.323 after version 1, and resulted in features like
FastStart and other procedures to reduce the startup traffic. There are
probably other examples of productive feedback between the two.
H.323 and ASN.1 eventually surpass a text based protocol because they are
better specified and can be implemented by compiler, securely, and
compatibily. A compiler handles more of the compatibility between
different implementations. Text based protocols are by definition vague.
Handwritten parsers offer more opportunities for bugs, incompatibilities,
and security problems. ASN.1 has other advantages that I won't go into
The most pointed past disadvantage of ASN.1 was that it was complicated
and expensive, because there were no free compilers, and decoders were
also hard to come by. That has all changed. Etherreal and other free
packet analyzers have H.323 modules, and can now easily incorporate other
ASN.1 based protocols. Recent enhancements to ASN.1 allows for
specification of encoder rules so protocols like IP or Q.931 can be
specified and implemented by compiler. There is really no reason for the
IETF not to consider using ASN.1 for protocol specification, and in fact,
it does in the GSSAPI.
On Thu, 21 Aug 2003, Dan Kolis wrote:
Since SIP is IETF not ITU its only reasonable to have internet believers
lean towards it.
H.323 ? Ahhh no thanks.
No serious look at these can even consider H.323 etc and its derivitives as
useful in the general case. The only reason they were used is the absence of
a better alternative.
Try to hook your Sony Playstation up through H.323
When would that move through committee? Spring of 2010 ??
Oh. Another reason for IETF to believe in it is that its basically a free
comm technology. H.323 wants to drag in the old timers and their costs
structures... dependance of geography, etc so there is a credible reason by