Covad Dips Toe into Metro Wi-Fi

Covad Communications is trying a different approach to metro Wi-Fi services, targeting services at businesses rather than consumers to create a revenue base.

Covad will build a Wi-Fi test network in a square-mile business district in San Carlos, Calif., using an approach far more narrow than what a regional nonprofit and a consortium including Cisco Systems and IBM had once envisioned. Covad also will limit its downside by agreeing to operate the network for a period of three months.

An earlier proposal by the Silicon Valley Network and Silicon Valley Metro Connect didn't take off, and revenue was the chief culprit. Azulstar, the startup that was to build and operate the network, couldn't get funding for two test networks at about $500,000 each.

That was supposed to be the start of a project serving 1,500 square miles and about 40 cities.

Technology really isn't the issue. Covad wants to find out whether it can get enough small business users to anchor a larger or more permanent effort.

Covad’s wireless business unit already serves business customers in San Carlos, allowing Covad to overlay the Wi-Fi capability on top of its fixed wireless broadband service. Central to the test is discovery of whether a repeatable financial and operational model exists.

Following the completion of the test, Covad Wireless will explore expanding the mesh service to additional locations in the region.

Covad Wireless operates California’s largest fixed broadband wireless network serving businesses, and the company views the trial as a way to test a theory: that it can reach incremental new customers in the very-small and home office segments it so far has not focused on.

Up to this point, Covad Wireless has focused on business customers requiring a T1 or higher bandwidth. So the issue is whether a sustainable business case exists for users who may not need, or are not willing to pay for, nailed up T1 connections.

P2P Issue is Hairy

P2P users amount to only one percent of Comcast's users, but those users are consuming 50 percent of the company's bandwidth, according to George Ou, an editor at large at ZDNet and the former technical director at TechRepublic.

Some say the ultimate solution is "more bandwidth." The problem is, that doesn't really address P2P bandwidth consumption, which automatically will scale to devour the additional bandwidth as well, some point out.

The problem is figuring out whether or not P2P applications really do pose outsize network loads, and must therefore be managed to preserve quality of experience for the 99 percent of other users, or whether P2P constitutes a lawful application that shouldn't be subject to rate shaping.

It doesn't immediately appear clear that there is a single answer. Traffic shaping isn't blocking. And even outright blocking sometimes is socially desirable. Blocking of viruses and spam come to mind. But even voice networks are designed to block some traffic when peak loads occur. One has to re-dial. Traffic shaping is somewhat analogous to the means used to cope with voice congestion at peak hours.

And, unfortunately, "more bandwidth" isn't by itself a solution to P2P network loading, it wouldn't appear. At some point, market mechanisms will have to be applied. At some point, the business pricing model will be applied in the consumer space: you can use as much as you want; you just have to pay for the usage.

Access Bandwidth: Size Doesn't Matter As Much As You Think

AT&T has been doing some testing recently. It bought cable modem service and then tested peak throughput and average throughput.

As you might have guessed, there's a huge difference. But the point is not to show that Digital Subscriber Line access is any better. It is to emphasize the point that access is only part of what determines end user quality of experience.

Every element of the delivery chain has to be optimized or quality of experience will be bounded by the weakest link. In fact, the tests suggest that real-world performance is precisely what users encounter.

Peak advertised speeds are possible at 3 a.m. During the evening hours, when home usage peaks, average throughput routinely drops as low as 300 to 400 kbps.

And that's just the difference between peak and average bandwidth. IP-delivered communications and entertainment also is subject to degradation because of latency and jitter, port contention and any number of other issues. In fact, port contention might in some ways be a bigger problem for mobile providers than raw bandwidth.

Mobile Broadband is Inevitable: History Will Repeat

About 22 percent of U.S. consumers go online wirelessly outside the home, compared with 16 percent of U.S. online households a year ago, says Sally Cohen, Forrester Research analyst.

Almost half of consumers Forrester surveys say they would like to do so.

Cohen says the growing interest is due in part to familiarity with home Wi-Fi networks as well as public hot spots. About a quarter of of consumers use Wi-Fi at home, she says.

The issue now is how fast mobile operators, Clearwire and Sprint can move to capture additional demand in the form of handheld and PC card forms of mobile Web access.

Because one thing is certain: history tends to repeat itself in the communications business. And that story is that services and features once considered "luxuries" become necessities, and therefore mass market products or even commodities.

Once upon a time families would gather around a phone at Christmas and make a long distance call across the country or world, at some point being exhorted to "keep it short." Once upon a time homes shared a party line.

The point is that broadband use has expanded pretty much as wired voice did. It was place based. At some point a small number of people started to use mobile voice. Now virtually everybody does.

The same thing is going to happen with broadband. People used to share bandwidth at work. Then they got service at home. The next wave will be mobile broadband used by people, just as mobile voice now is.

U.S. industrialists and entrepreneurs have been turning luxuries into everyday experiences and necessities has been going on since the 1870s, depictions of many as robber barons notwithstanding. As with many other innovations, the key is to systematize and standardize and wring cost out of the production of former luxuries so they can be provided as mass market necessities.

Mobile broadband isn't going to be any different. History does repeat.

Online Video Viewed by Half of U.S. Internet Users

eMarketer predicts that over half of the U.S. population will have watched video on the Web before the year is out.

By next year, more than 80 percent of all Internet users will have done so.

New Peak Load Issues for Mobile

There's only one problem worse than dealing with peak load, and that is average load that starts to look like peak load. In the voice world, peak load has been the bigger issue, not average load.
Data networks have peak load issues as well, but those issues are mostly about the number of bits to pushed through the pipe, not generally use of circuits or network elements or ports.

Wireless is starting to have other problems, though, as data usage grows. And you instinctively would think bandwidth has to be the issue. It isn't. But take the easy stuff first.

Just as there are two “rush hours” on the road (6 a.m. to 9 a.m. and 5 p.m. to 7 p.m.), enterprises typically experience two “rush hours” on their phone systems, says Art Yonemoto, owner of his own telecom expense auditing firm. For most enterprises those PBX rush hours happen at mid morning (10 a.m.) and early afternoon (2 p.m.). Hospitals, though, tend to have one of their busiest hours at 9 a.m., especially on Mondays, as their patients call to schedule appointments.

For the most part, though, enterprises handle their phone system peak loads quite well. Mobile networks cannot say the same. as they generally are designed to handle about 80 to 90 percent of calls on a non-blocking basis. At peak hours, blocking occurs.

Some calls simply go straight to your voice mail. The other obvious issue is a dropped call, which happens because you are moving from one area to the next, and the next cell tower has no free radio assets to hand off the call.

But there are other problems emerging, and that is amount of data traffic, and the different characteristics data applications impose on the network. Bandwidth alone is not the entire problem, any more than bandwidth is really the problem for voice traffic.

The issue is that radio resources are tied up even when not that much bandwidth is flowing over the radio network. An obvious example is a mobile virtual private network client, which essentially nails up a connection even when actual data is not flowing. The issue then is the strain on radio resources, not bandwidth as such. Other applications that don't actually consume much bandwidth might have lots of signaling and pinging. Some social networking applications and even mobile email devices can create that sort of stress.

So for wireless networks, it now appears application interactivity--not just bandwidth--is becoming a gating issue. It is an issue bandwidth alone does not fix. The additional new issue is occupation of radio resources.

Enterprise iPhone? Just talk to RIM Servers

Though there are other issues, Apple would get far down the road as an enterprise device if it did just one thing: ensure compatibility with Blackberry servers.

Though Microsoft Mobile is growing its share, BlackBerry is the device to beat. Apple will keep getting heat for its lack of security as well.

But the main thing is the ability of a user to get company email on an iPhone, not just on a Blackberry.