Nine facts about SIP [telecom]

They can be. To me, a trunk isn't a specific physical medium such as a loop start copper pair or a T1 demuxed from a fiber pair, it's a communications path that permits access to or from multiple endpoints such as a line pool for outgoing calls from multiple of handsets or an ISDN PRI supporting DIDs to those same handsets. If there's a physical characteristic that is common to trunks (as opposed to single lines as we might have at home or even Centrex deployments) it would be that it provisions a switch, not a handset. In that respect, SIP can be a trunk or not a trunk. (The phone just within my reach uses SIP in a non-trunk capacity; none of my clients currently use SIP trunking but it has been available from multiple telcos in this area for some time.)

I'll take their word for it, though telecom is chock full of examples of people figuring out how to get one medium through a path designed for another (usually voice): videophones, FAX machines, VoIP, etc. Of course, if you already have IP transport, then there are umpteen (posisbly better) protocols available to support a wide variety of other communication requirements... and those protocols might be more firewall-friendly than SIP.

Geez, my home POTS service provides local and long distance calling, I'm sure I could get a toll-free number for it, I've used a modem on it, and it's all billed on one invoice. Ooh, and if I had an old-fashioned phone it would light up during a blackout, too.

Most of the rest of these questions center not on SIP but on how [anything] over IP is not necessarily subject to the telcos' traditional limitations and costs. Many authors better than I have written about the implications of the way that the internet moves service differentiation to the edge of the network, as opposed to the traditional method of having a relatively simple device in front of the user and the services implemented in the network switches.

I'd be surprised if I'm the only reader here who has ever purchased fractional T1 voice service. Of course, there are fixed costs associated with the T1 that mean the costs don't scale linearly with the number of voice channels... but the same applies to anything-over-IP, where service good enough for one channel is probably good for a few, but when you approach that limit you need to take a leap up to the next tier of IP transport. Ironically, some of those tiers are derived from the same T1 (and its multiples) that the statement bemoans.

Actually, SIP was designed specifically with broader applications in mind; that's why it leaves the media layer to another protocol (RTP). In that it reflects the experience from SAP and from mid-nineties work on the "Integrated Services Internet" and multicast conferencing. Unfortunately, the presence of middleboxes leaves many of those designs undeployable. There are, of course, other protocols available, including most notably MGCP, but SIP seems to have won the day.

-GAWollman

SIP/RTP is a protocol which concentrates all mistakes that have been made in protocol design in the last 30 years into a single protocol. "30 years of experience, all ignored." This has brought crutches like STUN, makes security harder to do and does not do well with current network design.

Greetings Marc

--

-------------------------------------- !! No courtesy copies, please !! ----- Marc Haber | " Questions are the | Mailadresse im Header Mannheim, Germany | Beginning of Wisdom " |