VoIP at the University of Oregon The View from the Telephone Side Dave Barta, University of Oregon [email protected].

VoIP at the University of OregonThe View from the Telephone Side

Dave Barta, University of Oregon

[email protected]

VoIP ApplicationsTwo Distinct Evaluations

• Voip in the backbone– OPX replacement– Tandem networks

• Telephone switching systems to the desktop

OPX Replacement Locations

• Oregon Institute of Marine Biology (OIMB)– 100 miles from Eugene in Charleston, OR

• Pine Mountain Observatory– 200 miles east– Over a major mountain range– 19 miles from nearest paved road

• Future locations in local area

OPX Replacement - OIMB

• 2 analog ext. off UO PBX - 5 digit dialing

• Trunk group in OIMB key system– Dial Access Code (DAC) + 5 digits– Tail End Hop Off (TEHO) in OUS Cities– LD Access

• Rides US West frame relay T1– $450/month– Relatively low traffic

OPX Replacement - OIMB (cont.)

• Replaces one Off Premises Exchange line (OPX) @$250/month

• No signalling except supervision

• No compression (G.711)

• Quality accptable (but noticeable difference)

• 2 LEC service areas - GTE, USWest

OPX Replacement - Pine Mountain

• Like OIMB except two phones instead of key system.

• No OPX before - calls to campus LD. No access to dedicated LD services.

• Served by T1 frame relay circuit dropped from microwave tower on same mountaintop.

OPX Replacement - Local Remotes

• Rate changes creating advantageous climate– T1 @$300/mo.– DSL over 4-wire LADs @$48/mo., but

possibly going to $200/mo. – OPX @$31.50/mo., but possibly going to

$84/mo.

• In use at staff homes

• Contemplated for other off campus offices

VoIP to replace OPX is simple because:

• Usually no DID (although fairly simple with E&M tie trunk interfaces)

• No need for complex call accounting

• No or simple voicemail requirements

• Simple PBX translations

• Voice bandwidth requirements minimal

VoIP in the OUS Tandem Network

• Challenging

• Interesting

• Cost Effective (probably)

The Oregon University System voice network connects:

• 3 semi-large universities:– UO, OSU, PSU– All in the Willamette Valley

• 4 small regional universities:– WOU, SOU, EOU, OIT– Geographically remote

• OHSU - Public corporation in Portland

OUS ETN history and culture• 1988 - Contract with AT&T

– 8 PBXs - Sys 85-G2-G3– Dorm phones for LD resale– Copper and multimode fiber backbones– Voicemail - Audix - Intuity– ETN network and NOC

• 1992 - H.320 video and MCU• 1998 - Y2K upgrades to G3, Intuity, misc. systems• Intelcom Committee sets policy• Equivalent systems across OUS, with exceptions• Funding model subsidizes regional universities

OUS ETN consists of:

• Multiple ISDN-PRI hubbed to OSU

• AT&T SDN long distance from large universities w/ small schools using OSU SDN

• Video MCU at OSU - multipoint H.320 video or voice and w/ upgrade H.323 gateway and multipoint

• Shared CMS - ACD and trunking management statistics

• Fault reporting - receives system alarms, filters based on severity, and pages support staff

• NOC staff

• Call accounting

ETN features:

• 5-digit dialing between campuses - shared # plan• DCS - passes name info, answer and disconnect

supervision• SDN access for regional universities• Networked voicemail - name, # in message header• TEHO - Tail end hop off in each metro area• H.320 video between schools and to MCU• Reliability and stability

Forces of Change:• Cost Factors

– Long point-to-point T1s to EOU, SOU, OIT

– 384K H.320 video and voice traffic forced T1 #2

• Opportunity factors– Available OWEN bandwidth in backbone

– SDN rates dropping

– LD market competition

• Technological advances– Compression - Reliability

– Features (signalling) - Standards

ETN Replacement Options:

• ISDN-PRI SDN - Add SDN to regional schools and use it for LD, TEHO and (with QSIG) for all ETN functions

• VoIP to some or all schools w/ or w/o QSIG, depanding on vendor solution

• Hybrid Solution - VoIP in the Valley, ETN or SDN in the hinterlands

Major Assumption:

All existing features will be maintained or improved

Challenges• Bandwidth limited on frame circuits to regional

universities– At cheap frame price may be worth adding frame

circuits just for voice except…

– US West CIR unreliable

• Support for DCS signaling just emerging– Cisco to support some sort of signalling soon

– Lucent proprietary solution to packetize signaling information over IP network whether voice is IP or ETN

Other Issues

• Call accounting - on-net/off-net pricing based on PBX trunk group, so not a problem

• Video support– H.320 much easier with ETN network than SDN or 700

number option for schools without PRI trunking

– Move to H.323 will help but H.320 still growing

• Voice Quality - difference is noticeable, but meaningful?

Other Issues (cont.)

• Control– Telecom is sole owner and user of ETN network

(except for AT&T, US West, GTE, and PTI, of course)

– Using OWEN introduces many more interested parties and we would, for now, be the only isochronous app.

– SOEN/state politics - network ownership is a political issue which goes outside of higher ed.

• Complexity - more hardware, more protocols, more people, multiple alarm platforms

Other issues (cont. again)

• Installation cost– PBX cards– Router ports/cards– Possibly more SDN or frame circuits– Time and energy

• Remote maintenance– PBXs don’t support IP access for maintenance and

troubleshooting - modem or on-net EIA only– If/when IP maintenance access, security becomes an issue

Experimentation

• Cisco solution without signalling– T1 from PBX to Cisco router at UO and OSU– Discrete trunk group with manual DAC access– QOS available but not necessary– Voice great at G.711, noticeable difference at

G.729(a)– No fax or modem support– DID capability

Experimentation (cont.)

• Lucent Internet Telephony Server (ITS)– Outboard NT Server w/ T1 interface to PBX– DID but no DCS– Good voice, fax support– No compression– Ping and measure delay on call setup and give

busy if too much, but otherwise no QOS

Future Developments

• Cisco to support signaling– We’re experimenting

• Lucent– ITS replaced by inboard 3 slot card - G3V7– 3 slot card replaced by one slot card - G3V8.1– Support QOS via precedence bit - G3V8.2

VoIP to the DesktopWhat we have - What the users want

• Lucent Definity G3 V7 PBX – 7,000 stations, 550 PSTN trunks, – 5 T1 to AT&T for SDN, TSAA, 0+– 3 T1 to ETN network

• Features Supported– Extensive alarm reporting – Call coverage off-net and back– The usual pickup, park, transfer, + 6-way conf.

PBX features (cont.)

– ACD call centers w/ realtime, summary reports– Integrated and networked voicemail– Multiple sets from POTS to 36-button display– Supports wireless integration– Caller ID (# only) on digital sets and analog

(v8.2)– Legacy applications - alarms, modems, faxes,

emergency phones, elevator phones– ISDN-BRI

Experimenting with IP Phones• Purchased Selsius Trial Pack

– 6 sets used consistently on campus – Sets to remote locations

• One at OSU via Owen - worked fine• Took one to SLCC and used over congested

commodity internet - that was cool and useful• OSU to try same trick with telecommuter in Phoenix

- a potentially useful application.– Simple programming interface– Limited feature set– Lots of IP PBX rebooting– g.723(1) did not pass muster

Issues/obstacles to real implementation

• Full feature set, array of telephone sets, and set cost• Redundancy/reliability in controllers and core

routers• Distributed architecture means distributed backup

power - $ and space• e911 - solvable at the expense of mobility• Voice quality - compressed voice IS different, but

does it matter?

My opinion

• Seriously viable in small locations– Reduced feature requirements– Consolidated closets so power not an issue– Usually not fully integrated with campus

voicemail, etc. so not an issue– Price will decide

• Soon a good option on a campus without an existing cable plant

Last few opinions

• Wouldn’t want to be facing a PBX decision right now - cutting a large PBX is hard enough without it being a beta test

• PSTN response important– IP or some packet approach would reduce expensive

SONET gear

– It’s time for compression

– IP to PSTN breaks their pricing model to our advantage

• Hybrid of traditional telephony, IP, wireless