SIP in higher education

[(Dennis Baron)]

SIP.edu Conference Call May 11, 2006

*Attendees*

Dennis Baron, MIT
Chris Casswell, UNC Chapel Hill 
Joel Dunn, MCNC
Jeremy George, Yale
John Hess, Berkeley
Jeff Kuure, Internet 2
Dave Roeper, Independent Consultant
Christian Schlatter, UNC Chapel Hill
Mark Spencer, Indiana
Ben Teitelbaum, Internet 2
John Todd, Tello
Chris Trown, University of Oregon
Mike Van Norman, UCLA
Dave Zimmerman, Berkeley
Kurt Zoglmann, Kansas State University

*Discussion*

Today's call begins with Kurt asking about SIP routing and call
quality. He wants to know if they should allow calls to be routed
where they don't have as much control over the quality, or if they
should provide some way for users to choose how their call is routed
over either a PSTN or the Internet, perhaps by providing an access
number.

John Todd says that all of his company's traffic is routed over the
Internet. The have a system in place that monitors call patterns, and
if they see two calls to the same number from the same location within
a short period of time, they assume that the first call suffered from
a poor connection and the second call is routed over a PRI or a better
route. This does require some sort of state machine to keep track of
outgoing calls, but this isn't too difficult to implement.

Ben says that his belief with regards to the small IP telephony
installation that serves Internet 2 staff is that something that looks
like a phone number should live up to people's expectations with
regards to quality of service. Even if it is routed over the Internet,
they are more concerned about the quality. A SIP URI or ISN does not
look like a telephone number, so user expectations are different and
they are considered more experimental.

Dennis asks if Tello is doing E.164 lookups before routing calls. John
says that they perform parallel lookups on as many ENUM zones as
specified by the administrator for a particular site, with timeouts in
place for each database.

Kurt asks if anyone is using IPeerX. Dennis mentions that this is a
Pulver company, so when he does ISN testing from Freeworld Dialup he
believes that this uses IPeerX for call routing.

The majority of the call is devoted to Dennis talking about his
experiences with the latest version of eyeBeam, version 1.5.4 . This
version is available only for Windows, and is offered in both an
audio-only version as well as a full-featured version that includes
video and IM. A new license key is required for the full-featured
version, which Dennis had to request from CounterPath; he assumed that
this had to do with the purchasing arrangements at MIT.

Dennis experienced a variety of strange behaviors upon installation,
which was eventually traced back to a driver for his Eutectics USB
handset. He ended up removing all previous versions of eyeBeam before
installing the new version, which caused him to lose all of his
settings. According to eyeBeam, not all account settings are
transferred when performing an upgrade, so they recommend that you
save these before upgrading. The buddy list is handled differently in
the new version, so only the biggest group of contacts is transferred
during the upgrade. It is recommended that you merge all contacts into
one account in order to have them all transfer during the upgrade.

The biggest thing in new release is "zero touch" audio device
management. Previously, if eyeBeam was configured to use a USB device
which wasn't installed when you started eyeBeam, it would fall back to
the default audio devices. Now if you configure a device and start
eyeBeam without it connected, you will get a warning before using the
defaults. If you plug in the device it will allow you to change back
to the newly connected device.

There are also more codecs available, including H.264 support for
video, which Dennis has not tested thoroughly yet, mainly because he
does not know anyone else with H.264 capability. Christian mentions
that he has had issues with H.264, and has been unable to get it to
work.

There are also more wideband audio codecs, including Speex and
Broadvoice. The Telechemy code is used to do MOS scoring as well. It
has TLS support for secure SIP and secure RTP, as well as sets of
security options that Dennis has not really played with yet. In theory
you would be able to set eyeBeam to accept only secure connections or
offer only secure connections, for example.  There is also support for
on-the-fly codec changes, QOS, and more support for USB audio devices.

In the new version, account-specific settings are now split from the
overall options, and dial plans are supported, which is a feature that
Dennis uses on his Snom phones. There are also improvements in IM
support and presence, but Dennis does not know if there is any sort of
manual override for presence settings. Finally, audio-only calls can
now be recored as WAV files; previously both audio and video calls
were recorded as AVIs.

Dennis is asked about the QOS support, and says that it will do
Diffserv, DSCP, and 802.11e. The MOS scoring will tell you what the
upstream and downstream bandwith usage is, what codec you're using,
and gives a MOS score, but provides no information about
signaling. Dennis would like to do some TLS testing, but has noticed
that in the past when he used a Snom phone to call Harvard it tried to
SRTP but failed.

Other changes include the removal of the diagnostic window, which
Dennis thought was one of the better tools for SIP debugging. It has
been replaced with a tool that generates comma-separated value files
which can be opened in Excel, but he has found nothing that provides
good SIP logs. Dennis feels that the product has been somewhat "dumbed
down" and does not seem to expect SIP experts as customers, which is
not necessarily bad but can be problematic if you are knowledgeable
and want to investigate problems. There is no way to set codec
preferences or video framerate, and the audio and video tuning wizard
is missing which Dennis found useful for testing before making a
call. Additionally, you cannot set the presence for each account;
Dennis suspects that eyeBeam now automatically detects a presence
server at the other end of a call.

Dennis is asked about using his Eutectics cordless handset with
eyeBeam. He did not know that a cordless version existed.

Someone asks about using VOIP over a satellite Internet
connection. Ben mentions that there has been some testing at Texas A&M
using Tachyon geostationary satellites, and recommends contacting Walt
Magnusson. Dennis says that Art Gaylord at WHOI may use something
similar to contact research vessels at sea, and believes that
something similar is being done in Alaska.

John Hess asks if anyone has any experience with OpenSER and RFC 4244
History Information, which is a SIP-specific event notification set to
replace CC-Diversion. Berkeley has bought an Avaya Communications
Manager and is looking to integrate it with their Interactive
Intelligence CommunitM-CM-) messaging platform. The Avaya signals call
forwarding using 4244, and the CommunitM-CM-) expects to receive a
CC-Diversion header. They are looking to translate this
information. Nobody on the call is familiar with this.

The next call will be in two weeks, on May 25th. Cisco will be doing a
presentation, and by their request the call will be open to Internet2
university members only. Interested participants will need to RSVP, as
there will be a different conference bridge number.