SIP in higher education

[(Dennis Baron)]

SIP.edu Conference Call October 20, 2005

*Attendees*

Dennis Baron, MIT
Ed Guy, Pulver
Candace Holman, Harvard
Jerry Keith, UC Riverside
Jeff Kuure, Internet 2
Mark Spencer, Indiana
Ben Teitelbaum, Internet 2
Mike Van Norman, UCLA
Garrett Yoshimi, Hawaii

*Discussion*

Today's call begins with Dennis discussing MIT's progress with regards
to replacing their current Pingtel SIPXchange system with SER. They
are also migrating to new Cisco gateways, while trying to maintain both
systems in parallel during the change. Overall, things are going well,
but they've encountered a known bug in the Cisco software.

In terms of ISN development, freenum.org currently has an ISN route server
with a few entries in it. It is  running Asterisk, and Ben has been
working on a SER script to do ISN parsing and ENUM-like lookups. There
have been discussions about bringing up a beta test root, which MIT has
offered to run. This service creates many questions regarding nomenclature
and control of such a system; Dennis is trying to avoid offending anyone
by creating something which could be viewed as being only for educational
institutions or US residents.

Following this, Ben discusses the new additions to the cookbook, which
is currently under construction. The pieces exist but are still being
formatted and copy edited. The draft version is available online at
http://internet2.edu/sip.edu/isn/.

Ben has also posted some recipes for use with SER, as tested on the
internet2.edu domain. This domain runs SER for approximately 100 users,
using SIP URIs as the primary identities. Incoming invites first look for
registered user agents; if none are found, an alias lookup is performed
and the call is passed to the person's desk phone. There is a nightly
database dump from the staff directory to the SER MySQL table, which also
inserts a rather cryptic ISN alias to the SIP URI. Prior to this, there
was no numeric dialing plan, but outbound PSTN calls were supported. From
registered users, calls come in looking like 7, 10, or 11 digits at the
domain being served. They are disambiguated at the server by length -
4, 5, or 6 digit numbers are ITAD calls, while 8, 9, or 10 digits are
PSTN calls.

Ben wrote a Perl program that does ISN resolution; this is at the bottom
of the recipes. For outbound calls, ISN calls must be identified. There
are regular expressions to strip local prefixes if they are used;
since Ben had no existing numeric plan he didn't need to use a local
prefix. From his user's phones, they can simply dial the ITAD directly. He
looks for numbers * numbers and uses the exec deset method to route the
call appropriately. For inbound dialing, he looks for numbers which are
4-6 digits long, and looks them up in the alias table.

After Ben's explanation, Ed from Pulver mentions that the ISN/ITAD dialing
is up and running. The prefix is **192, followed by the user number,
followed by a * plus the ITAD number. Inbound dialing works already; there
are still some political issues with using prefixes for outbound calling.

Ben mentions that all numbers in the state of Oregon's government are
now reachable by ISN.

The remainder of the call centers around discussion of caller ID and SIP
dialing.  There are various suggestions for handling the identification
of remote callers from SIP devices, and how this identification is handled
throughout various PBXs and proxies during the call. One possible solution
focuses on using the Cisco RPID field for passing along caller ID or
ITAD identification along with the call. Questions are raised about what
characters are valid in an RPID, how inbound and outbound proxies could
insert values into the RPID, and if multiple values are allowable. More
research into the RPID is required.

The next call will be on Thursday, November 3.