Transcript Performance Management

Voice-Enabled Applications: Implications for
Service in the Converged Network
Tom Schmidt and Joe Schmid
March 2002
People’s Choice
Best of Show
Investor’s Choice
Agenda
 Introductions.
 Brief company overview. (We’re not in sales!)
 Technology trends.
 Where does the value in tomorrow’s “perfect”
network reside?
 Examples of decomposing today’s Voice
Applications.
 BeVocal’s value proposition.
 Challenges of creating a voice services framework
in a converged world.
2
Why does BeVocal exist?
 Improvement in speech reco.
 Commodity, standard hardware platform
availability.
 Internet model of standard software platforms
and interfaces. (VoiceXML, HTTP,etc.)
3
Why NOT the Internet model for voice?
 Stability
 QoS
 No clear billing/revenue model
 Very different expense points and models
 User experience requirements/expectations
 Heavily regulated
 Sloth-like carrier movement
4
About BeVocal
 Founded in March 1999, based in Santa Clara, CA
 Headcount – 100+
 Raised $70M in venture funding to date
– Mayfield Fund, US Venture Partners, Technology Crossover
Ventures, Trans Cosmos USA, prominent angel investors
 Launched BeVocal consumer service offering in June
2000.
 Carrier-focused with proven in-network deployments
with tier-1 carriers: BellSouth, Cingular, Qwest.
5
What we do
Your network is your biggest asset.
BeVocal’s voice software helps you
generate more money from your
network, by delivering more value and
creating a better experience for your
customers.
6
Delivering bottom-line results to carriers
 Enhanced voice services enable telecom companies to
 Achieve 10-25% increase in subscriber revenues by
– capturing additional monthly subscription fees of $5-10/sub/month
– up-selling subscribers to premium bundles
– attracting new subscribers via unique, differentiated offerings
– improving customer loyalty with highly personalized, branded services
 Achieve 30-40% decrease in service costs by
– automating call centers for customer support and directory assistance
– consolidating voice applications onto an open, flexible voice platform
7
Delivering value to consumers and businesses
 Simply by speaking, callers can instantly connect to people,
businesses, and information from any phone
 Enhanced voice services include:
– Voice Dialing – dial by saying a number or name in your address book
– Voice E-mail – read and reply to e-mail messages via phone
– Voice Mail – play and send messages using voice instead of a keypad
– Voice Portals – access the best of the web over the phone
 location and travel services (driving directions, business finder,
weather, traffic, and flight info)
 information services (stock quotes, news, sports)
 entertainment services (movies, TV dramas, horoscopes, lottery)
– and much more …
8
Delivering enhanced services in a new way
Low risk deployments with fast time to market in a
cost-effective and scalable manner.
•
•
Vendor-independent technologies based on open
standards (VoiceXML)
Incremental, pay-as-you-grow build-outs
•
Assets reusable across business units
•
Assets reusable across products
•
Services created and designed by carriers and third-party
developers
Make your network more valuable in helping to defend
and grow market share in the face of competitive and
regulatory challenges.
9
Proven deployments with network operators
 Wireless carrier customers
 Wireline carrier customers
 Telematics service provider customers
10
Major Trend in Telecom Networks – Stupidity

1997 David Isenberg (at AT&T) writes “The Rise of the Stupid
Network” comparing the network models of the PSTN and the
Internet.

PSTN model:

–
Intelligence embedded in the middle of the network. (AIN)
–
Assumes circuit switched voice is the bulk of traffic.
–
Assumes of scarcity of computing resources.
–
Centralized control.
Internet model:
–
“Dumb” transport in the middle.
–
Pushes intelligence to endpoints. (Prof. Katz’s “Hourglass” diagram.)
–
Smart endpoints drive the network, not limited by design assumptions
of the network itself.

Many implications to each model, primarily centered around
innovation.

Stupid network likely to “win” in the end, but continued coexistence is likely for some time.
11
Characteristic of Two New Protocols – Stupidity

SIP and SOAP embrace the model of “dumb” transport
by:
– Standardizing basic protocol plumbing, but…
– Allowing for extensible headers, opening up…
– The ability to layer additional services on top of this basic
plumbing.
– Again, intelligence has been pushed to the endpoints.

Extensibility examples:
– Standards for IM and Presence (SIP for Instant Messaging
and Presence Leveraging Extensions – SIMPLE WG) already
being layered over basic protocol as extensions in SIP
headers.
– Security, Digital Signatures (W3C SOAP Note) already being
layered over basic protocol as extensions in SOAP headers.
– Nothing violated or changed in basic protocols.
– Intermediaries (SIP Proxies, SOAP Intermediaries) can
process messages regardless of extended headers.
12
A Key Question

“Where is the edge of the network?”
– In today’s PSTN?
– In today’s WAP-enabled wireless networks?
– In today’s TCP/IP internets?
– In tomorrow’s (today in some places…) 3G wireless networks?
– Answer is dictated in part by where new services can be
added.

Likely answer will be…
– Edge of the network is on my intelligent wireless devices –
with always on, high-bandwidth packet network wireless
connectivity.

In evaluating evolving network designs, try to identify
“edge” of the network.
13
Network edges – VoiceXML vs. SALT

VoiceXML (http://www.voicexml.org)
– Brings the web model (HTTP + standard mark-up) to voice
services, but…
– Still embraces the PSTN-style “intelligence in the network”
model, rendering audio UI (VUI) onto dumb terminal phones.
– Both evolutionary and revolutionary. Fits into existing telecom
networks, but opens up service creation by using web model.

SALT (http://www.saltforum.org)
– Mark-up standard for specifying true multi-modal, interruptible
interaction with users.
– Accounts for PSTN-style, voice-only deployment scenarios,
but…
– Big win is when SALT interpreter executes on intelligent
devices. Back to intelligent endpoints model.
– Allows innovation on endpoints – intelligent devices + backend services.
14
The Network Paradox (http://netparadox.com)
 "The perfect network is perfectly
plain, and perfectly extensible. That
means it is also the perfect capital
repellant, [which] implies a
guaranteed loss to network
operators, but a boon to the services
on the 'ends'." -- Roxanne Googin in the
September 2001 issue of her “High Tech
Observer” report.
 “Perfect” in terms of:
– Delivering bits in large quantities, at high
rates.
– Open to new, innovative services at the
endpoints.
15
Decomposing Monolithic Voice Services

Today’s Voicemail:
– Sits as “intelligent peripheral” on the SS7 network.
– Delivers many functions in one monolithic black box:


Renders UI onto device – today’s “dumb terminal” 64kbs audioonly phones.

Contains UI logic, controls user interaction.

Message store.
Decomposition under way from an unlikely source –
Microsoft:
– .NET My Services moves “message store” into network cloud.
– Available anywhere, anytime, on any device.
– Decomposed model looks very different, very flexible.
– Who owns the billing relationship? Keep an eye on this one…
16
Another decomposed example: VAD

Voice Activated Dialing where:
– Subscriber can provision phone to use any provider of Voice
Activated Dialing on the network.
– Subscriber’s personal/business Contacts loaded by VAD
application from web service.
– After recognizing destination party, uses Presence web service
to detect location of called party, intelligently routing call or
offering options to caller.

Implications:
– Applications become more like “assemblies of services” rather
than monolithic black boxes.
– Framework for assembling services is needed (and valuable).
– QoS, SLAs, response latency become critical among these
decomposed pieces.

Side Note: One off-the-wall prediction about the LNP
equivalent in the consumer-oriented web services
world!
17
BeVocal’s Value Proposition
 Provide a software platform that allows
creation, deployment, and management of
innovative voice [and multi-modal] services on
a large scale. (Think BEA WebLogic for
voice/multi-modal apps.)
 Requirements of such a platform:
– Enable “assemblies of services.”
– Event-driven, interruptible.
– Scalable, fault-tolerant, etc.
– Does the “heavy lifting” of building innovative apps.
18
BeVocal Foundation Platform (Part One)
 Addresses common problems of building voice
apps:
– Prompt and grammar management.
– Dialog flow, templates, building blocks.
– “Universals”, context sensitive help.
– Versioning.
– “Hot” deployment and upgrade (across sites).
– Application health and monitoring.
– Usage and performance statistics.
– Assembling combinations of apps (“portlets”) into
“portals”.
19
BeVocal Foundation Platform (Part Two)
 Offers compelling services needed to build
carrier-grade voice applications:
– Alarming and monitoring.
– Flexible billing infrastructure.
– User Profile and Personalization.
– Address Book.
– Content Feeds. (Weather, Traffic, Sports, etc.)
– Security (Authentication and Authorization).
– Service Discovery and Registration.
– Distributed, real-time session data.
20
Enabling Unique Multi-Modal Applications

When fully deployed, (SIP-enabled smart endpoints)
SIP will open up a variety of innovative applications.

BeVocal’s Platform should be prepared to take
advantage of these unique abilities once available:
– Not everything is a phone call! Use generic “trigger” concept.
– Build “interruptability” into service creation environment and
make it easy.
– Support distributed, shared session state.
– Account for multiple “legs” of sessions (conferencing, chat,
etc.)
– Account for “long running” sessions. (e.g. Notification of bid
status on eBay, etc.)
21
Challenges
 Design and implement an architecture that:
– Fulfills today’s needs for creating, deploying, and
managing voice services in a carrier environment.
– Makes it easy for third party developers to create a
variety of apps.
– Allows for compelling services provided both by
BeVocal and “assembled” from third parties.
– Will deliver on tomorrow’s opportunity to easily
develop interesting multi-modal applications.
– All done in a way that is manageable and costeffective on a carrier-size scale, with performance
that meets mass-market criteria.
22
Q&A
Thank you for having us!
23