Web Hosting and the Central Office

Web hosting is a service generally provided by specialized ISPs, not by local broadband access
providers. I know of no instance of a broadband wireless network providing this service,
though it is certainly feasible. Web hosting requires what is known as a server farm, which is
just what the name implies—a data center holding numbers of servers supporting numbers of
Web sites and Web-based broadcasts, multicasts, unicasts, and transactions. Often, though
not always, the Web hosting company will form a relationship with a specialized provider of
transport such as NaviSite, which will ensure the expeditious delivery of hosted content.
Web hosting is a complicated business that is really separate from the high-speed Internet
business. As with broadcast video, it demands a fundamentally different kind of physical plant
(although the two could be combined, albeit at considerable expense). Web hosting is best left
to specialists.
The MMDS broadband wireless services began as networks dedicated to delivery of video,
and some local networks still function in this manner. A few have even been successful, primarily
in remote areas where a well-established cable incumbent is absent, although today with
the ready availability of direct satellite broadcast services almost everywhere it is doubtful how
long the traditional wireless cable model will remain viable.
Also, Multichannel Multipoint Distribution Services (MMDS) convergent services networks
have been set up on the cable television model where a couple of 6MHz channels have
been allocated to residential high-speed access while the rest of the spectrum has been
assigned to television channels (though I know of no outstanding successes based on this
approach). Given the relatively limited amount of bandwidth available in the MMDS bands,
the model seems somewhat dubious, though it could work with digital television equipment
where several channels apiece are assigned to the 6MHz frequency slots. The problem is that
most such networks have been built on the macrocell model and cannot reuse the data channels,
and therefore no considerable population of data users could be served.  My belief is that the wireless cable model is wrong for most markets today. In any case, the
limited bandwidth with which to work would appear to confront the network operator with a
crucial choice: to specialize in one-way video programming or to specialize in two-way highspeed
data. The two approaches require different equipment complements within the central
office. The equipment requirements for data services are discussed at length throughout this
chapter and do not differ significantly in terms of higher-layer networking equipment from
what one would find in a DSL network. A video-oriented system, on the other hand, requires
what is essentially television station hardware, including video tuners, up and down converters,
satellite feeds, ad insertion servers, scramblers, video archival hardware, and so on. Given
the poor record of wireless cable plants in making profits, I do not see why anyone in a developed
market would try to resurrect this approach.
I have alluded earlier in Chapter 3 to the possibility of doing video on demand (VOD) over
a wireless broadband network. No one to my knowledge has done so, but techniques have
been developed for VDSL networks that could easily be applied in the wireless realm and
would not require the reallocation of vast amounts of bandwidth. Such an approach would
require the purchase of a dedicated video server, a device that may range in price from
thousands to tens of thousands of dollars depending on port count, and, if continuous programming
were provided, banks of tuners that subscribers could access to select individual
programs. I would hasten to add that at this time VOD is a completely unproven service offering
for wireless broadband, but the possibility exists that it may find a market in the future.
Somewhat related to VOD in the sense that VOD is essentially one of its subsets is interactive
television. Again, this represents a market that is embryonic in the wireless realm but that
could conceivably emerge in the future and certainly bears watching.
Interactive television consists of programming that invites the direct participation of the
viewer. Examples include the following:
• Participating in t-commerce, where a viewer can click an item in an infomercial or even
an episodic program and then call up a Web site with product information and the
capacity for credit card transactions
• Participating in game shows such as typing answers to a question or voting
• Selecting viewing angles during a sporting broadcast
• Ordering takeout food off an on-screen menu
• Summoning video catalogs containing desired products
• Calling up news clips of events of interest
Interactive television applications have enjoyed some success in European cable television
networks but have been generally rejected within the American market to date. Currently
no specialized equipment exists for their delivery over wireless broadband; within cable networks,
specialized set-top boxes and headend servers are required to enable such services.
There does not appear to be any pent-up demand for such services, but the market could
change.
Successful network operators have to be able to anticipate changes in markets, and for this
reason close monitoring of interactive television is indicated at present. Clearly, any early entry
into this market with improvised equipment is not advisable, though.