Tuesday, December 10, 2002

Rethinking Telecom Andrew Chapman, Narad Networks Mike McCue, TellMe Michael Stumm, SOMA Networks   Andrew: Building ethernet switched networks on HFC.  The network shouldnt be so stupid you cant do something with it. Mike: Call Center, Directory Assistance, Comm Functions.  Michael: Broadband wireless DSL alternative in the last mile.  (Last mile $500-1500 per sub via copper)   Kevin: why is there any hope at all for any of you?   MM: Telecom is a business, where there is change there is opportunity.  ILECs needing to invest. AB: Huge industry, we are the 3rd company I did with my co-founder in this space.  Everyone uses the network.  Has a bet on www.longbet.org that one of the ILECs will go bankrupt in 2007. CATV is the Rodney Dangerfield of the industry, but have change alot.  Twice as many subs as PSTN for IP.  Cox $250m/yr in enterprise businesses.  Looking to grow their business, do new things and will buy on a good ROI model.  MS: 12-16 month ROI breakeven and it will be deployed.   Look for where deployments are happening.  They made a good shift from CLECs to ILECs which are buying.  Soma offers 12Mbps, synchronous but shared and distance sensitive.   Kevin: US Broadband vs. Rest of World? MM: Korea has greatest penetration of broadband, but that's under the current definition.  If you really have broadband you never worry about bandwidth.  We build a fully switched gig Ethernet service.  This audience is stuck on a narrowband platform.  SIP with unlimited bandwidth, distributed end points managing services.  Net needs to know priority and economic rational. VoD experiments will not scale (compression barriers, server), need CES or IT end products MS: Putting all the services on the end users requires cooperation of all sorts.  Most end users have certain kinds of software that they cannot manage.  Firewalls and content filters are moving further into the network.   MM: Give up on best efforts networks.  Fully capable QoS network that can read and prioritize enable rich bandwidth.  AB: You have to have a balance between De/centralization.  Its not just complete decentralization.  (Centralized management, decentralized infrastructure). MM: The network's awareness of classes of service is a central function.  TWC wants to offer disaster recovery services in NY, but could partner with IBM to do so -- a balance of De/centralization.  You have to allow people to make people to make money and invest when they think they can.  You have to find a way to allow people to share in the profits.  Doc: Cox forcing him to buy Asymetrical? (nice consumer question, yawn) MS: Flood the network, then shut it off to terrorize your nieghbors AB: We dont sell to humans.  SMB   Kevin: What will change in terms of applications? MM: Call center with voiceXML $3-5min to $.03-.10 per minute.  Directory with more services available at 411 through automation.  Voice activated dialing, email -- 5-10 from now getting a voice prompt instead of dial tone (I guess dial tone isnt God after all). AB: Migration of enterprise applications to SMB.  Depending upon network based storage.  Video enriched email, storage of personal video, video conferencing.. MS: More managed as services firewall, content filtering, conference calls (ahh...remember TimeShift?).  AB: Tools need to be created to allow users to create their own applications in P2P fashion.  Like Blogging.   Kevin: Incumbents fighting it? MM: They wont change AB: Cable winning.  More apps that are PC-like will complement MM:  Blogging like tool using voice using TellMe  http://studio.tellme.com  Kevin: Who pays for the minutes? MM: Phone billing infrastructure is incumbent opportunity --  (bah, not at when billing costs are 20% of revenues) MS: Competition drives action.  Either from small greenfield deployments or large guy gets aggressive MM: SMS integration with Voice Messaging Kevin: Web Services? AB: Doesnt have anything to do with the web, it does with services.  Doing things at Layer 7 that are interesting and people want to pay for. Kevin forsees a movement in OSS towards a Web Service componentized architecture MM: telecom guys want to install eBay in their plant, they dont get web services AB: metro ethernet deployment becuase they support the applications people want to consume.  CATV more open, serving these apps and will be more dominant. AB: QoS is not anti-consumer, but its crappy because of the network its built on.  Managing flows through networks marries willingness to pay with a service -- virtuous cycle. AB: Amphone will be created to allow Universal Service to die. MS: Reg mandated services are expensive because of the old network they have to run on MM: Work with wireless then shift to wireline to avoid regulatory issues 7:17:13 PM    comment [] David Isenberg Note: David doesnt use the word commodity, I stuck it in below Kevin: In my dreams I would come up with a simple idea like Davids that is so powerful and everyone gets. Im shocked to have never gotten below layer 7 in the discussion today and have people still call it infrastructure.  That's not infrastructure.  Infrastructure is important and uncertain.  Today's news: Massive failure by PTTs and ILECs, getting worse, and I can't wait. Distance is dead, except where its being held back by regulation and monopoly Yahoo BB hapan giving away modems, 12Mbps/ ~0.5 upstream 12000Yen/mo=~$15USD, started in April, took to September to get first 1M customers.  5M ADSL subs in an 80M person countries.  120k Fiber to the home (FTTH) customers.  Growth.  Cost of customer acquisition make him cringe The Future (not evenly distributed yet) VON in Atlanta, small company, Global IP Sound.  Shrink wrapped bundled with a Compaq Pocket PC with vanilla 802.11 with telephony that sounded better than PSTN. WITH NO TELCO IN THE LOOP. When the net is better than POTS, the cash cow goes away SIP (Session Initiation Protocol): what HTTP did for documents, SIP will do for communications POTS: Sure you can do Internet on it, you can do it with smoke signals too Simple networks provide connections, not services.  Providing commodity connectivity.  Services are enabled by smart edge devices. Gilder's Law outpacing Moores Law, Depreciation iinversely nverse, Engineering effort scales according to nobody's law. End-to-End principle, 25-30 years old, similar to stupid network -- if you can do something at the ends or at the middle, do it at the ends to preserve your options, use of the network will be different in the future. Internetworking shifts control from network owner to end user Telephony yet another application,  Bob Cannon and Bob Pepper at FCC Winning apps not created by telcos, most haven't been discovered yet (SIP and presence management!!) Old biz model Voice-monthly income network subsidized physical subsid Stupid model apps are services -- a vibrant market network is protocol -- a commons physical a not subsidized -- he says the big question is the business model? -- the commodity business Who owns and runs the network? Telephone company? Difficult transition to the horizontal model Cable company? Difficult to give up old video entertainment model Municipalties? 125 experimenting Utilities? New kinds of company? Customers? Politics of End-to-End...big list of ideals vs. the Dark Side GOLIATH LOST Q&A I posed the utility model question and he says Im right, but there are alternatives (he is right too) Server should be at the edge of the network 7:16:23 PM    comment []

© Copyright 2003 Ross Mayfield. Last update: 1/6/2003; 8:06:17 PM. This theme is based on the SoundWaves (blue) Manila theme.