1. Video Development Initiative
H.323
Jill Gemmill
University of Alabama at Birmingham
Doug Pearson
Indiana University, Bloomington
Tyler Johnson
University of North Carolina, Chapel Hill
3. Desktop & Room Systems
• USB or Appliance
• Affordable
• User Friendly
4. What is H.323 ?
An ITU-T standard for bi-directional
exchange of voice, video, and data
Applies to an IP network
H.323 is a set of standards for group
communication
TCP Call setup & control
UDP for audio/video
6. H.323 Video Standards
Video is optional; H.261 required
H.261 Picture Size
– QCIF (176x44 pixels) is required
– CIF (352x288 pixels) optional
H.261 Compressed Data Rate
– 64kbs – 1.9 kbs
H.263 SQCIF, 4CIF, up to 16CIF
(1408x1152 pixels)
7. Point-to-Point & Multipoint
Point-to-Point
Dial by IP address ! or alias
user user
Gatekeeper
user
user
MCU
user Multipoint user
Register with Gatekeeper
Connect through MCU
8. What is ViDe.Net?
An international virtual network providing
video teleconferencing, telephone and
collaboration services over advanced
networks. Architecture used in I2 Commons
A voluntary collaboration and open forum
A mesh of interconnected 75+ H.323 zones
Zone: a collection of users administered by
the site.
RESULT: individual campuses and network
providers interconnect, creating a seamless
global environment for teleconferencing and
collaboration.
9. Network Requirements
Please use switched Ethernet
Category 5 wiring is part of the above
10Mb/sec should be adequate for end
points
Much higher bandwidth at MCU
(multiple 100Mb/sec cards in some
systems)
10. Duplex Mismatch: the
“Silent Performance Killer”
A connection set for auto-negotiation,
failing to see auto-negotiation at the
other end, sets itself to the default –
half-duplex.
Auto-negotiation doesn’t always work,
even when both sides are set to auto
Auto-negotiation occurs repeatedly at
intervals; what’s right the first time
can be wrong later
11. Detecting Duplex Mismatch
Show switch port stats; if
mismatched:
– High CRC or Alignment errors at full
duplex end
– Late collisions at half duplex end
UAB sets all user ports to 10/half by
default
12. Firewalls
H.323 uses these IP ports:
– Statically-assigned TCP ports 1718 –
1720 and 1731 for call setup and control.
– Dynamically-assigned UDP ports in the
range of 1024 – 65535 for video and
audio data streams.
Firewalls don’t allow unrestricted
ports: typical modern firewalls and
H.323 don’t get along so well.
13. Firewalls – Solutions for H.323
• [bad; non-scaleable] Allow unrestricted ports
for specific, known, external IP-addresses.
• [better, but still not so good] Use feature of
some videoconferencing clients to confine
dynamic ports to a specific, narrow range.
• [OK, but extra admin work and cost] Use an
H.323 application proxy.
• [best] Use a firewall that snoops on the H.323
call set-up channels (static ports) and opens
ports for the audio/video (dynamic ports) as
needed.
14. Network Issues
Ongoing, undiagnosed problems with
H.323
– Jerky video over uncongested networks;
why?
– Quality of H.320 vs. H.323 over
uncongested networks
– >.1% packet loss = unacceptable audio
(ITU)
– >.5% renders session unusable
16. H.323 Network Requirements:
Jitter
Definition: Variation in latency over
time.
For switched end-points, the primary
source of jitter is variation in the
store-and-forward time, resulting from
network load.
H.323 is very intolerant of jitter;
clients reduce resolution
Rule of thumb?
17. H.323 Network Requirements:
Packet Loss
Typically due to router or link
congestion
– >0.1% packet loss (ITU) = unacceptable
audio
– >0.5% renders session unusable
19. Tools: ViDe.Net Scout
Scout is a web-based distributed
network performance analysis tool
developed at the University of North
Carolina at Chapel Hill.
Scout makes use of the NetIQ Chariot
performance testing engine
http://scout.video.unc.edu/
21. Scouting Out Problems
Public Health Outreach Project
THROUGHPUT
Remote Health Clinic
connected to Internet2
via xDSL
Original diagnosis was
h.323 problem
ISP refused problem
ownership until
presented with Scout
results
PACKET LOSS
22. Iperf – Example on Intercampus DS3
http://dast.nlanr.net/Projects/Iperf/
MRTG utilization graph
showed bandwidth peaking
at capacity ~ 10:00am –
2:00pm
As utilization peaked on
the DS3, jitter measured by
Iperf rose to unacceptable
level
Iperf also reported periodic
high packet loss, with no
apparent correlation to the
low-resolution MRTG
utilization reports
23. The End-to-End Problem
Problem: Poor video and/or audio in
a conference between campuses
Need: Timely, useful assistance
If there’s a firewall, it could take
weeks!
24. Suggested Solutions
Articulate the E-2-E problem to network
management and engineers
Bring all engineers together in a place and
time to share information.
Establish and use reliable communication
tools
Improve diagnostic tools
Have good network documentation for all
networks
25. Future ViDe.Net Direction:
Middleware for Video
Directories : key enabler of video
teleconferencing.
Directories : management tool for tracking
and supporting users
Directory : a portal for account requests,
support, user information updates, etc.
Goal: Directory lookup, authenticated click
& dial
26. Middleware for Video: NSF
National Middleware Initiative
NMI Release 1 due this Spring
h323Identify and h323Zone object classes
will enable h.323 attributes to be added to
campus LDAP directories
Coordination with Internet2 Middleware
activities will create a globally searchable
directory of video users
Will allow automatic user account
management by integration with enterprise
directories
28. SURA/ViDe Digital Video
Workshop April 23-25, 2002
4th Annual Workshop
Birmingham, Alabama UAB Campus
Digital Video Applications: collaborative
conferencing, streaming video, and
storing/serving video-on-demand
Technical Training & Discussion
Program/Technical/Vendor Advisory
Committee – please volunteer.