The meeting began at 7:07 PM at the Terman library with introductions by Imsong Lee.
Roland Alden of Internet Fiber presented background information on his company and the approach he is taking towards supplying fiber Internet service. At this time they are not actually providing the service, but are negotiating with several developers to do so.
Economic issues for service on fiber connections were described first. Base cost of Internet access now is $20/month which people willingly pay. The capital cost behind this dial up product is about $17/month i.e., about 1/15 the cost of the dialup port. The cost of any type of faster infrastructure based on copper wire or coax cable to the user is significantly higher. Assuming that people won't pay more than $40/month for faster service, the economics don't make sense since capital and maintenance costs are higher than $40. It requires significantly higher than $40 to $50 in monthly charges to make a profit.
Telcos have done cost analyses and decided fiber to the home is cheaper to maintain than fiber to the curb because the electronic box for distribution to groups of homes is expensive and has to be maintained. Problem with fiber to the home is the very high relative capital cost for initial installation. The only justification for fiber to the home is the greater bandwidth. Is it needed? What is the actual amount of bandwidth needed? Answer is that the users need as much bandwidth as they can get. Just as people want bigger hard drives, more RAM and faster CPUs, they want more bandwidth. The requirement for any network architecture is the aggregation of customers. Central offices aggregate many customers together and offer significant economies. Long distance data transmittal would be via SONET, medium distance data transmittal the alternate is ATM vs. Ethernet. Ethernet is widely used now, mainly for local service and internal data handling. A high bandwidth physical circuit supports different services and changes the types of services supported over time. Highest efficiency in the system is obtained by making changes with software, not physically. Ownership and competition among service providers for high bandwidth service is problematic. Cost of rebuilding legacy installations is huge, which inhibits telcos and cable companies. Drops, in-home wire cost, and getting wires from the street to the home greatly increases cost. About 53% of cable interconnection cost is labor. Fiber connectors for service outdoors are more expensive, $30 vs. $10 for indoor connectors. CTE mismatch of glass fibers in a system causes problems because the fibers move relative to other components, causing a loss of signal quality. Fibers and the connectors work well inside, but less well outside. Cost of running fiber lines rises greatly in residential neighborhoods because of all the connections and equipment needed. Most technology now is derived from relatively long distance loops - 15,000 feet is common because that's historically what was needed.
Stan Smith asked how the PacBell system in San Jose that recently was abandoned fits in with the model being discussed. It was fiber to nodes, with DC run into the home to run the phones.
Alden wasn't familiar with the system in detail, but it probably was a standard ATM system. Phone service requires powering the phone so that 911 calls can be made even if all power goes off. This requires a separate power supply for groups of homes to provide emergency power for phones. It's a big issue and can increase costs significantly. ATM is relatively expensive compared to Ethernet. All types of service require some kind of transition from the backbone to the home or other user. Getting the best combination of fiber backbone, connections to the home, and internal fiber wiring is important.
Fiber can provide enough capacity to deliver all of the services, but it has lots of legal complications which would be difficult and expensive to resolve. How do different providers share costs of the fiber? Who is responsible for repairs and maintenance? What happens if a service is interrupted? If there is a common owner of the fiber it is simpler but there still are problems. Owners of the fiber or wire usually don't want to let others use the facility. When Bell was broken up the government made the phone company give internal wiring to the building owners. That also made building owners responsible for maintenance and repairs for internal wiring.
Internet Fiber has the strategy which says that network plant to the home is crummy, and they can do better. New residential construction uses the same obsolete approach that has been used for decades. Cost to run copper wires to the home is about $300 to $1500/home. Coax cable cost is about the same. Fiber costs would be higher, but not much higher. Doing it better by installing fiber service during construction adds value to the residence. If there were no phone wires in the house, what would the cost difference be? If ADSL is great, why do offices use Ethernet instead? It's cheaper and faster than DSL or cable. Sales were $10 billion for Ethernet cards and ports last year. Ethernet equipment is relatively cheap, and it can support runs of several thousand feet. Using Ethernet eliminates many of the costly transitions.
Van Heimke asked why ISPs are reluctant to use Ethernet connections, and how are the transitions made to Internet backbone? The preference for ISPs is to stay with T1, T3 or DS3 rather than single mode fiber WANs.
Alden said partly it was heritage, partly cost. It's hard for telcos to track use and costs with Ethernet. The slowest Ethernet is in the 5 to 10 Meg range which is more than most users need or would pay for..
He is selling the neighborhood network, as an amenity for new home buyers. There will be Internet jacks all over the house for direct hookups. Ethernet ports with 2 fibers cost $170 for a 4-port hub but they have limited distance capability, maybe 2 KM. Cost is .25 to .75% of total housing unit cost for adding a high capacity connection to the Internet. Having the fiber Internet connection built in will add significant value to the home, just as undergrounding of power lines or cable does. Phone companies can have limited capability compared to potential demand for high bandwidth services. Apartment builders are more interested in the opportunity than are single-family developers. It's important to convince developers and buyers that having the fiber Internet hookups adds real value.
Locally if fiber goes right past the house you still can't just hook up. (Residents in Southgate asked MFS for hookups after the MFS fiber to the Research Park was installed, but were refused because they have no interest in residential users at this time.) High speed access to the Internet and a local loop which offers real competition from multiple service providers are very desirable. Attracting competitors requires significant aggregation. Persistence (24 hour availability) is more important than bandwidth. Over time there is no limit to what bandwidth people will buy at some price point.
Imsong asked what kind of infrastructure would be installed in a new neighborhood by the Internet Fiber. It would be Ethernet to every home with T1 lines, one router facing the neighborhood, and track actual usage to be sure that people get the speeds they need. Adding caching on the LAN speeds up service greatly. Serving less than 500 users can be uneconomical. If many people get on at the same time the data transfer rate slows down noticeably. Switched Ethernet is less vulnerable to slowdown from many simultaneous users than is hybrid fiber coax.
Van asked about throttling. It would be done for the local loop, and can be done in several places. Given the bandwidth available on fiber, the homes can have phone service also, and can get it from people other than the local telco. Problem with providing it is that phone service is regulated and has entry issues.
If the city owns the fiber it can avoid lots of the costs for switches and many regulatory problems. City ownership eliminates many of the problems and inefficiencies.
Joe Villareal asked why not also use wireless for the phone? Roland Alden said it isn't as good as wired systems, and if a cell phone is used for access the costs are significant because of the per-minute charges..
Marvin Lee asked who would be more likely to use the features, since the model is designed for newly built homes, but locally almost all houses are built already.
Telcos have been trying to sell people video phones for years without success. The thing people want is Internet. It's possible to take phone business away from PacBell, but it isn't a desirable market because it's hard to get people to switch. Retrofitting homes can cost up to $5000, but that will greatly improve property values. Being close to the Internet greatly improves desirability of the area and raises property values. Right now almost 1/2 the worldwide Internet traffic goes through Palo Alto and the DEC facility downtown.
If other areas put in their own fiber infrastructure they could draw off business and income from Palo Alto. Palo Alto wants to make it more desirable to connect to the Internet here and avoid losing users to other areas with a lower cost of living. Make high speed Internet access available at a price, and many people here will be willing and able to pay it and get the service, and thus increase home values. Over time the high speed Internet connections will raise property values. Having some people pay to upgrade to Internet access and others not doing it will create a split in housing values. The added cost of the wiring is small compared to the total house value. Eventually homes will be put on fiber before they change hands. It will work here if it works anywhere.
Mary Jo Levy asked what is a reasonable aggregate to be economical? Eight or 16 houses served from a distribution box is economical now. Since Internet service costs $20/month now, how can it be made economical to spend all the added money for fiber and high speed access? It's far cheaper to use fiber than ISDN or ASDL connections, and having fiber internally adds long term value to the home. If the cost of fiber connections was added when areas are undergrounded the incremental cost is small and people would be very willing to pay it. Existing City loan programs for undergrounding power, phones and cable to the home allow a 10 year payoff at low interest rates via the utility bill. This could be extended to include financing fiber installations.
Van said the existing City fiber loop is within 2 km or less of 90% of all homes in the city. That may be too far in some cases. Underground plant does fill with water, so that has to be taken into account in design and construction. The best electronic switching equipment is designed for indoors. Putting it outdoors requires building local protective boxes, and if they are too big or intrusive people complain.
Michael Silverton asked if the City should look for a single provider, different providers for specific areas, or something else?
If one entity owns the infrastructure and can give any user a share on the facility, then with existing technology only one provider can do the service economically, or it takes much more expensive technology. There have been requests to allow more than one organization to provide service. In practice only one organization provides the service efficiently, and if they fail to perform a new provider can be brought in. It would be a negative sales feature to offer people a variety of service providers. Most people would not want to make the choice. Also offering the ability to chose would run costs up. Radical choice is difficult to offer economically.
A cost comparison table was presented:
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
Bob Moss noted that his street has been undergrounded for years, and house prices haven't been significantly different from non-undergrounded streets. Value added to property varies with the type of addition. For example, a new bath recovers more of the cost in added value than new carpets. It's unclear how much of the cost of fiber installation would be recovered in higher house value.
There was a comment that people wanted the bigger hard drives, more RAM and faster CPUs because the cost has been going down so fast it is very affordable to get them. Unless the cost for bandwidth also falls, most people will be less willing to pay for more speed on the Internet.
Imsong ended the formal discussion at 9:10 PM. Participants were asked to continue the discussion separately if they wished to.
Next meeting will be Wed. Feb. 4, 7:30 AM at Terman Library.