From The Archives: e-xpedient - 100 Mbps Internet Access, $100/month

Note to readers - I'm having surgery this week and in my absence, I'm featuring some of my favorite articles and columns from my decade of writing about Broadband Wireless Internet Access. e-xpedient / CAVU was the best example I've ever seen of an "Isenberg Stupid Network" Broadband Internet Service Provider. Yes, they certainly had their problems... but technologically, and the business model, e-xpedient was a marvel. e-xpedient's Miami network was a real marvel, combining Free Space Optical and 60 GHz systems for near-Gigabit, 5-9's reliability, license-exempt inter-building links. No one else that I'm aware of has even come close to the elegance of e-xpedient's business model. The "full" e-xpedient story has yet to be written, and I hope to be the one to do it. So many stories, so little time... Lastly, the e-xpedient in this story has [apparently] nothing to do with the current company that bought e-xpedient's assets; apparently they renounced using wireless as a bad idea.

Boardwatch Magazine
Feature Article by Steve Stroh
Submitted for February, 2001 issue

e-xpedient - 100 Mbps Internet Access, $100/month

Brian Andrew is in the business of creating bandwidth junkies. Andrew's company, CAVU, Inc. offers the e-xpedient service (www.e-xpedient.com), and gets customers hooked on high speed Internet access, and then they can't do business without it.

CAVU gets customers addicted to very fast Internet access by offering 100 Mbps Internet access to every tenant in the building - from the four person not-for-profit trade association, to the multi-floor prestigious law firm in multi-tenant buildings in downtown core areas. What gets potential customers interested in the e-xpedient service is that their usual Internet tasks - web browsing, accessing ASP web pages, is so much faster. The fact that the Internet connection is operating at 100 Mbps is nearly meaningless for many customers. The customers simply understand that "doing Internet stuff" that used to be really s l o w is now so fast they think that something might be wrong, only to find that their file download is completed, or the email message with the very large attachment was sent in a few seconds rather than tens of minutes.

What converts a prospective customer to paying customers is e-xpedient's pricing. e-xpedient offers 100 Mbps Internet access at prices beginning as low as $100 / month.

CAVU certainly isn't alone in targeting this market segment. In high-rise buildings (high-density, multiple business units - MBUs) there are numerous competitors, including Allied-Riser, Broadband Office, yipes, Cogent, Teligent, Winstar, Advanced Radio Telecom, XO Communications, etc.

CAVU's edge is in the combination of performance and pricing, and in support of the pricing, strong execution behind the scenes. Some of the aforementioned competitors offer 100 Mbps... at price points beginning at $1000. Others competitors offer pricing approaching CAVU's,  for "T-1" connections (1.5 Mbps). CAVU seems to be alone in their particular price/performance niche.

CAVU's Secret Weapon

How CAVU can achieve this price performance properly begins at one of CAVU's key suppliers, Triton Network Systems (www.triton-network.com). Triton manufactures the "Invisible Fiber" line of microwave radios, and the feature set and price performance of this product is a key part of how CAVU can offer e-xpedient service and, most importantly, do so profitably.

Brian Andrew was one of the founders of Triton Network Systems. Andrew founded Triton with the idea that 100 Mbps was going to be the next big leap in Internet access, and that fiber, though plentiful and reasonably priced in certain buildings and areas, simply would not be ubiquitous. Triton's business plan was to provide high speed, easy-to-install microwave radio products to bridge the gap between fiber-rich buildings and fiber-poor buildings. Triton planned to sell Invisible Fiber radios to ILECs, CLECs, and other service providers. In general, the market Triton had envisioned didn't materialize- the most frequent comment was "We don't need a 100 Mbps radio - we need a cheap T-1 radio because we can sell a lot of T-1 circuits". Andrew, and the Triton Board of Directors were disappointed at the slow uptake of the Invisible Fiber product line, and eventually Andrew and others left Triton to found CAVU, Inc. and begin offering the e-xpedient 100 Mbps Internet service based on the use of Triton's Invisible Fiber products.

The Invisible Fiber product line is an incredibly sophisticated product. The network architecture of Invisible Fiber is "consecutive point" - a series of point-to-point links that compose a counter-rotating ring. If any single radio fails, data is already flowing in both directions around the ring, and since there is a second radio on each building, users should not notice a radio failure (other than some reduction in performance since bandwidth from each direction is used).

The two main features of Invisible Fiber that distinguish it from competition is that it is very simple to set up, and it allows for one-to-one frequency reuse. Both of these features play heavily in CAVU's execution. Simple to set up means that very few "High Priests of RF" are needed, and a typical installation building startup requires only familiarity with the Invisible Fiber radio. One-to-one frequency reuse means that only one microwave channel in an entire metropolitan area is needed, as opposed to large swathes of spectrum that several competitors claim as a competitive advantage. To obtain spectrum for e-xpedient, CAVU only needed to license a single microwave channel in each market.

CAVU's Execution

CAVU's business plan is basically to deliver 100 Mbps Internet service to customers at a price point competitors are unlikely to be able to (profitably) match. CAVU's entire corporate system is geared for fast deployment and cost-effective infrastructure.

An e-xpedient market begins with selection of a Point of Presence (POP). A POP is somewhat centrally located, with good fiber connectivity, with the ability to host up to twelve radios on or near its roof (six "rings"). For the most part, CAVU is agnostic about its selection of Internet backbone provider, although for purposes of redundancy and route diversity, two different backbone providers are used at each POP (each one supplying different ends of a ring).

After a building has been surveyed by an advance crew, and an engineering plan written (to, among other things, to coordinate IP addresses and do some RF planning), a well-developed "pick list" is sent to CAVU's warehouse. There, every component necessary for a building installation is picked, configured, tested, and put on a pallet and shrink-wrapped. For a building other than the point-of-presence (POP), this process takes about an hour. The level of detail is surprising - for example, cabling (other than Category-5 Ethernet cable) is pre-cut and pre-terminated to minimize installation time at the site. All IP addresses are set up in the warehouse. Even POP equipment - several rack's worth of equipment is pre-built, pre-tested, and pre-configured in Orlando and shipped as single units, again, to minimize setup time.

Locally, when a building is to be "lit up", a local electrical contractor is hired to install mounting plates for the radios on or near the roof (the radios can also be installed on walls) and install conduit runs from floor to floor. On floors where there are active components, electrical circuits are also installed. Typically, this process takes three days. CAVU's installation team takes over at that point. A small crew installs the radios, pulls cable, and installs the electronics. Typically this process takes only one day. In well under one workweek, a new building is online.

CAVU's presence on most floors of a "lit" building" is a small junction box with six Category-5 Unshielded Twisted Pair RJ-45, 100baseT Ethernet jacks. All that's required for a new customer to be connected is a simple run of Cat-5 cable to the telco closet on that floor, a few seconds spent crimping an RJ-45 jack on the cable, and the customer is (physically) connected.

Once connected, a new customer trying to browse the web is redirected to a CAVU web page where they can enter their billing information. A customer is prompted for a credit card number or a (electronically processed) check. Payment is for a month of service in advance. CAVU claims not to "bill" in the most conventional sense; CAVU doesn't have a "department of opening and stuffing envelopes".

Once a customer is connected and operational, they become the responsibility of CAVU's Network Operation Center (NOC). Another departure from convention is that CAVU does not have a customer service department. If a customer has a problem, their call comes to NOC personnel. CAVU's NOC personnel can monitor the entire network, to an unusual level of detail. All devices in the system are monitored with SNMP, including such mundane things as temperature of equipment rooms, backup battery voltage, and whether a cabinet has been opened. CAVU is also making increasing use of Webcams to monitor its facilities.

Again, this process is very rapid for a building in an established market. CAVU's local marketing and sales departments are identifying the likely prospects, both potential customers and potential building owner management (occasionally making use of the former's need for cost-effective Internet service to persuade the latter). CAVU has already established relationships with several electrical contractors to do the "heavy" work of floor-to-floor conduits and the work on the roof for the radios. CAVU's local technical support team is well established. When a new building is added, the building setup process is repeated, with one difference. An existing ring (CAVU's standard practice is that up to ten buildings can be on one ring) is expanded during a late evening maintenance window. Typically, rings are planned so that only one radio on two existing buildings have to be re-aimed to bring a new building into an existing ring. Only one radio is taken re-pointed at a time so that the service is not disrupted. Again, new customers in an e-xpedient-equipped building require only a Cat-5 cable to the telco room to begin using e-xpedient, very rarely requiring e-xpedient technical personnel to come on site. (Ironically, one of CAVU's minor problems at the moment is that customers are asking for help in locating low-cost firewall appliances that operate at e-xpedient's 100 Mbps)

The Customer Experience

When a building is "lit" (and burned-in / tested for a week or so), an e-xpedient sales team sets up a kiosk in the lobby of the building. All tenants are individually invited to come down and sample the e-xpedient service. Tenants sometimes have trouble believing that "available now" means exactly that.

As much as the customer enjoys the speed of the e-xpedient service, even more noticeable is the reliability. It's always on (assuming the bill gets paid regularly, and if not, service automatically ceases, and is restored instantly once payment is made). CAVU goes to great lengths in its system design and operations to insure Five Nine's reliability (99.999 uptime - telco grade reliability). For customers used to dealing with dialup, ISDN, and DSL (connectivity solutions they could afford), the reliability is remarkable. The speed of the connection quickly becomes addictive, often causing offices to upgrade computers and network connections to remove the "bottleneck" of 10 Mbps Ethernet hubs and network cards.

The e-xpedient Ecosystem

It takes some time to understand that CAVU's e-xpedient service is a complex "ecosystem" with all the parts dependent on the other parts. The speed of the e-xpedient service gets customer's attention, and the pricing and reliable service keeps the customers. The biggest difference between CAVU and its competitors seems to be that CAVU provide 100 Mbps Internet access profitably, and repeatably. Some observations:

• The use of the Invisible Fiber radios means that a reliable high speed network can be constructed and expanded quickly and inexpensively (compared to fiber or other wireless systems)
• The extensive preparation in CAVU's warehouse means that installations can be completed quickly
• That electricians are the only "external" personnel needed for installations keeps costs low
• The simple infrastructure (inside the buildings) means that new customers can be added quickly and inexpensively
• The radios don't require extensive RF expertise to install, so relatively rare (and expensive) RF skills are largely unneeded. The skills that are needed - router configuration, TCP/IP are relatively available from numerous industry training programs
• The promise of stock options extending to all employees, and interesting, intensive work and a rapidly growing company with room for advancement translates to high motivation of all personnel
• No billing / collections department means fewer personnel on payroll, and very few bad debts to deal with
• That deployments can be accomplished quickly means that competitors are scrambling to catch up.
• In 2000 and beyond, high speed (let alone 100 Mbps), reliable Internet access for a reasonable price is a relatively easy sale - one of the few barriers to converting a potential customer is existing long-term contracts with other providers.
• CAVU appears to have enough cash already committed from venture investors to activate a number of new markets and claims to go cash-flow positive very early in a new-market activation.

It's equally clear that CAVU has built itself to scale rapidly:

• The warehouse crew ably demonstrated their ability to ship many "buildings" per day
• The warehouse has a large "surge" capability, easily capable of supplying the startup of a new market with a few week's notice (mostly needed for the buildup of POP racks)
• CAVU has an extensive training system in place and is graduating new personnel every week.
• CAVU's management personnel clearly understand long term support issues such as standards, practices, and documentation very well
• Standardized procedures are evident, down to the careful labeling of each component to be installed in a building with an SNMP-compliant label
• CAVU's first NOC in Orlando is clearly capable of managing far more markets than are currently deployed, and a redundant NOC is already under construction
• CAVU's regional management structure is already largely in place for national deployment.

IP Play / Wireless Play

CAVU is a "pure IP" play. CAVU sells only high-speed IP connectivity. Although CAVU does not offer hosting or application services, it does supply bandwidth to hosting companies such as Application Service Providers (but will not sell bandwidth for redistribution, for example, to an ISP). With its high speed network, CAVU is ideally positioned to take advantage of current and next-generation services based on IP such as VOIP and Virtual Private Networks (VPNs), videoconferencing / remote training, as well as large file transfers, especially critical to legal groups (a common complaint there is "don't email me the document, it'll take to long - it'll be faster to fax it to me"). Voice Over IP (VOIP) works very well on the e-xpedient service, which CAVU uses to great advantage through the use of VOIP phones for its internal operations including its field operations.

To some extent, CAVU is a "Wireless Internet play". It does use wireless as its infrastructure and in doing so completely avoids the delays and expenses of fiber deployment. Using licensed spectrum it avoids the complexities of license-exempt spectrum, but in requiring only a single channel per market translates to a very low cost for use of licensed spectrum. That CAVU uses wireless to connect buildings is totally transparent to its customers. That CAVU uses RF, rather than optical is a significant reliability factor. Although optical systems are indeed promising, their reliability will have to be proven, especially in climates with significant precipitation, over long periods of time. In the meantime, RF technology is improving rapidly in efficiency, performance, and cost.

There are a number of proprietary aspects to CAVU's system that investors look for in deciding whether there are sufficient barriers to competition. The prospect of a competitor spending several years developing an infrastructure to compete with CAVU is daunting. CAVU's extensive experience with deploying the Invisible Fiber radios will serve it well. CAVU claims to have obtained sufficient spectrum commitments to enter any market it targets, and very favorable arrangements with a number of Internet fiber companies sufficient to support rapid buildouts in numerous markets.

Potential competitors will no doubt note CAVU's successes with the Invisible Fiber radios and being buying them and deploying them also (despite many of CAVU's personnel being founders of Triton Network Systems, there is, apparently, no special relationship between the two companies other than supplier and customer).

Indeed, the most potent argument against any effective competition to e-xpedient, even with the use of Invisible Fiber radios, is that most of CAVU's potential competition simply doesn't now, and likely cannot (profitably) offer a competitive service - in time:

• Several of CAVU's competitors provide telephony services - conventional, switched telephony services requiring a (expensive) telephone switch, payment of termination fees, inefficient (compared to IP) use of network, a need for skilled personnel, etc. This at a time when the price of telephony services is being driven to zero by the use of Voice Over IP and related services. CAVU does not offer voice services, and the e-xpedient service is ideal for using VOIP.
• Other CAVU competitors make use of older microwave systems. One competitor uses radios that require 4 channels for each building link - one channel each for transmit and receive for the primary, and a backup system that uses two other channels. Not to mention that the radios in question don't have active power control, and so these channels cannot be easily reused.
• Fiber-based service providers are potentially potent competition until the incredible costs of running fiber to a new building are factored in. The cost of that fiber installation has to be amortized, and in the current financial climate, relatively quickly, requiring high service fees. e-xpedient service is installed in buildings with existing fiber-based service providers and winning customers.
• Most damning (amusing... sad...) are the CAVU competitors that will persist in trying to sell T-1 equivalent (not to mention DSL-equivalent) services in the face of e-xpedient 100 Mbps services at nearly the same price points.

Tomorrow, the Continent. The Rest Of The World... Soon

It's tough to see what, if anything, will slow CAVU. The one thing that will govern the rate of deployment to new markets for the e-xpedient service is capital. Given that CAVU offers a very desirable product that customers are proven to buy if they possibly can and CAVU has created a scalable infrastructure to deliver that product, the only missing ingredient appears to be more capital. The obvious answer is an IPO... but market conditions are hardly optimum enough for that. So, CAVU continues to seek enlightened investors to accelerate their buildout.

The merits of what CAVU is offering now are compelling enough. What CAVU could potentially offer in the near future with only slight deviation from its current plan could make CAVU even more compelling (and these ideas are, for the moment, speculation):

• Deployment of wireless LAN services in served buildings. No "hassles" with wiring. The new 802.11a Wireless LAN standard will offer 50 Mbps and faster using 5 GHz
• Begin servicing high-density residential buildings with 100 Mbps service
• Extension of service ("spurs" off of a ring) to single-customer buildings (perhaps with the customer paying a one-time charge for the radios and installation)
• In conjunction with a VOIP service provider to handle the "back end", VOIP telephony services
• Extension of services into smaller sub-markets such as suburbs using other wireless technologies, for example to provide sub-100 Mbps service (but substantially faster than T-1) to a small office park (where it's not cost-effective to deploy Cat-5 cable to each unit).

Remember those innovative brackets that support the Invisible Fiber radios? One of their best features is that all the "aiming" is done on the bracket, not the radio. So when a radio needs to be swapped out, the replacement radio doesn't need to be re-aimed; it's pretty much ready to go. So, when a particular building needs more bandwidth (for example, a hosting company moved in on multiple floors), all that needs to be done is to replace the Invisible Fiber radio with an OC-12 Invisible Fiber radio, operating at 622 Mbps. All the infrastructure is pretty much in place. Perhaps a new fiber might need to be pulled, but there's conduit between the switch/router and the radio. Perhaps the switch/router needs to be upgraded, but that's a job of all of thirty minutes. For a total installation time of perhaps 2 hours, OC-12 service is now available. It's going to be very, very hard for the competition, any competition, to catch up.

Perhaps more than any other Internet Service Provider, CAVU is capable of delivering the real potential of the Internet to most mainstream businesses - very high speed Internet access, at reasonable prices.

Think that CAVU has aimed high with their plans? You'd be right. One clue is in the breakout of the CAVU acronym, well known to aviators as "Ceiling And Visibility Unlimited".

Footnote:

In the article above, I frequently use the term "microwave". In a strict, technical sense, that description is inaccurate, since the spectrum(s) in question are more accurately located described as "millimeter wave" (a reference to the wavelength of the spectrum in question). I found the term "millimeter wave" a bit too clumsy for use in the story, and chose to use the term microwave, which I felt would be better understood by the audience of this article.

(Sidebar)

e-xpedient's Secret Weapon

If CAVU has a "secret weapon" for making the e-xpedient service work, it would be the Invisible Fiber line of microwave radios manufactured by Triton Network Systems. It helps to understand that, immediately prior to CAVU, Brian Andrew was one of the founders of Triton Network Systems. It's no coincidence that the e-xpedient service looks a lot like what Andrew expected the Invisible Fiber product line to be used for.

Some of the original gallium arsenide semiconductor microwave technology for Invisible Fiber came from Lockheed Martin, who originally developed it for missile guidance systems. Triton then spent tens of millions of dollars commercializing the technology for use in communications. Gallium arsenide semiconductor technology is incredibly tricky to "tame"; yields are well under 50%, and there is significant variation on each individual part. For example, each gallium arsenide device must be temperature characterized, and the variances in performance over temperature must be factored into the operation of the radio.

But, the product that results is absolutely state-of-the-art, not only high performance but as close to Plug and Play as is possible for a microwave radio.

Most of the "magic" of Invisible Fiber comes from its incredible dynamic power control, which is important for a number of reasons. The most important is that if the radio is transmitting just enough power to reach the other radio, then that spectrum can be reused in as little as a few miles. If the radio is transmitting too much power, reuse is not possible, the radios would interfere with each other. Yet, it's not enough to only transmit at low power levels - sometimes you need more power, like in a rainstorm, or a longer "shot" between buildings. The radios can vary their transmit power, automatically, from picowatts (billionths of a watt) to a full two watts.

Even fully automatic, dynamic power control isn't enough. Weather conditions vary widely. In Seattle, for example, weather patterns (relevant to microwave communications) change slowly - fog and rain rolls in gradually. In Orlando, for example, weather patterns change rapidly - an intense rainstorm comes up quickly. If the radio "over-corrects" - ramps the transmit power up and down too slowly (to compensate for the Orlando example), communications could be temporarily lost. If transmit power is ramped up too quickly (to compensate for the Seattle example), then the radios could interfere with each other.

One of the parameters programmed into the radios is their latitude and longitude (and that of their partner radio). From this, the radio then derives two critical pieces of information. The first is the distance between the two radios, and approximately how much transmit power is needed, initially, to reach their partner radio. The second piece of information is what "rain region" the radio is located in (again, the Orlando versus Seattle climates) and therefore how aggressive the transmit power level should be changed (Seattle = gradual, Orlando = aggressive).

It's instructive that Triton chose to derive this information in the radios themselves. A more usual procedure would have been to provide the initial power settings, and "step-change aggressiveness" as parameters to be set manually, after consulting some tables in the installation manual, etc. Instead, the latitude and longitude can be easily determined by GPS receiver, and there's more than enough computational power in the radio itself to derive the required information.

The dynamic power control and other elements such as Forward Error Correction and the counter-rotating ring architecture all contribute to the radio's 99.999% (5-9's) reliability and a Bit Error Rate (BER) of  10 to the 12th - for all intents and purposes, fiber quality.

The rest of the "magic" of Invisible Fiber is that the radios relatively easy to set up and maintain. For one, they're in one piece. That means there's no need to get deeply into the "mechanics" of setting up antennas, waveguides, attenuators, etc.; hardware that's tough to set up and requires experienced (and expensive, and increasingly hard to find) personnel to do right.

For another, Triton came up with a pretty amazing mounting bracket that removes much of the hardest work of aligning a radio. The bracket is, literally, a masterpiece of mechanical engineering in its own right (it's patented) allowing an installer to easily align one radio with another with only a screwdriver and a digital multimeter. All of the alignment between two radios is done in the brackets, not the radios themselves. Once the bracket is locked down into proper alignment, the radio can be replaced, and the new radio "inherits" the alignment.

Once the radios are "on net", either via a RF connection to another radio, or via its fiber connection, then they can be managed and checked out remotely. Each radio has two fiber connections - one for data, and the other for connecting to other radios on the same building. The "radio" fiber connects directly between the radios (via a patch panel at the same location as the power supply for the radios). The purpose of the "radio" fiber is that the radios have an "out-of-band" management channel - a bit of bandwidth reserved strictly for management and control. Even in the rare event that the data channel is completely saturated, the radios can still be managed and monitored via the management channel. The radios also perform Layer 2 routing - if data coming in one radio is destined for elsewhere in the ring, the radio simply passes the data along to the other radio for transmission to the ultimate destination in the ring without involving the switch/router in the building.

(sidebar)

e-xpedient Markets

• Birmingham, AL (date not specified)
• Charlotte, NC (date not specified)
• Cleveland, OH
• Columbus, OH (date not specified)
• Jacksonville, FL
• Little Rock, AR (date not specified)
• Los Angeles (North), CA (date not specified)
• Oklahoma City, OK
• Orlando, FL (Winter 2000)
• Richmond, VA (date not specified)
• Rochester, NY (date not specified)
• Salt Lake City, UT

(sidebar)

e-xpedient Pricing

Service Level: e-xpedient DE
Typically replaces: Dialup modem
Included GB / Month: 1
Price / Month: $100

Service Level: e-xpedient XE1
Typically replaces: DSL
Included GB / Month: 2
Price / Month: $250

Service Level: e-xpedient XE2
Typically replaces: Fractional T-1
Included GB / Month: 10
Price / Month: $500

Service Level: e-xpedient TE
Typically replaces: Full T-1 / Frame Relay
Included GB / Month: 20
Price / Month: $1,000

Service Level: e-xpedient VIPER1
Typically replaces: Multiple T-1's
Included GB / Month: 50
Price / Month: $2,500

Service Level: e-xpedient VIPER2
Typically replaces: T-3, ASP / Hosting
Included GB / Month: 100
Price / Month: $5,500

By Steve Stroh

This article is Copyright © 2001, 2007 by Steve Stroh.

Comments

[Jonathan Wells]

Steve,
Your article really made me chuckle! With CAVU and Triton long dead, FSO dying a long drawn-out death, and 60 GHz proven viable for just a few hundred meters, I wonder how this ever got off the ground. The business model migh be good, but the technology was way overhyped and flawed.
You might want to note what is happening today in Las Vegas (www.1velocity.com) and San Francisco (www.wiline.com) - the same business model, but with far more robust 70/80 GHz technology .
Hope the surgery goes well!
Jonathan
[email protected]