"Surprising" AT&T Stance on Net Neutrality?

Some people might be shocked to learn that AT&T complies with existing Federal Communications Commission rules. Some people might be shocked to learn that AT&T actually already agrees that "best effort" Internet services ought to treat every packet the same as every other.

“We use the principle of ‘us on us,’” says AT&T  CTO John Donovan. “If we take an external developer and ourselves, we should not be advantaged in how long it takes or how much expertise is required."

"I don’t think it needs to be that complicated," he says. Does any application run by any third party work as well on the network as an AT&T-provided application?

"Outside applications need to be on an equal footing with our own applications," Donovan says.

But that's part of the problem with net neutrality. It is very hard to define and covers a range of business discrimination issues, network management and performance practices as well as potential future services that consumers might very well want to buy, that provide value precisely because they allow users to specify which of their applications take priority when the network is congested.

As a working definition, net neutrality is the idea that ISPs cannot "discriminate" between packets based on the owner or sender of packets, or on the type of lawful application, or block lawful packets.

The latter principle already applies to fixed broadband access connections, and the new change might be the extension of such rules to wireless providers. What is "new" in the current net neutrality debate is that concept that no packet can be afforded expedited handling, compared to another.

At some level, this is common sense. One wouldn't want video packets or voice packets sold by a third party to be disadvantaged, compared to video packets sold by the Internet access provider, for example.

But that isn't the issue in the current round of discussions and the possible FCC rulemaking. The issue is more an issue of  whether "affirmative" packet handling, as opposed to "negative" packet handling, will be lawful in the future.

"Negative" packet handling is sort of a "thou shalt not" approach: application providers should have a reasonable expectation that their best-effort Internet traffic will be handled the same way as any other application provider's traffic is treated. So ISPs "shalt not" provide any quality-of-experience advantage for their own application bits, as compared to any other bits delivered over the network.

All that sounds fair and reasonable, and in fact ISPs (after a few notable cases of interference), have concluded it is not worth the public outrage to block or delay any packets to heavy users, even when networks are congested, for the purpose of maintaining overall user experience for all the other users.

But there are several issues here. Good public policy would forbid business discrimination, a situation where any ISP could attempt to favor its own applications over those provided by its competitors. Back in the "old days," an example might have been a refusal by one telephone company to deliver calls from a rival.

But the network neutrality debate is far more complicated than that. There is a broad area where network management policies designed to maintain performance might be construed as business discrimination, even when the purpose is simply to protect 95 percent of users from heavy demand created by five percent of users.

Under heavy load, real-time applications such as video and voice suffer the most. So either end users might want, or ISPs might prefer, to give priority to those sorts of applications, at peak load, and slow down packets less sensitive to delay.

The problem with crudely-crafted net neutrality rules is that they might make illegal such efforts to maintain overall network performance for most applications and most users. One can hope that will not be the result, but it remains a danger.

The other issue is creation of new services or applications that can take advantage of expedited handling. Users might want their video or voice packets to have highest priority when there is network congestion. Crude net neutrality rules might make that impossible. But one can hope policymakers will take that sort of thing into consideration.

Net neutrality is a very-complicated issue with multiple facets. Ironically, end users might, in some cases, actually want packet discrimination.

More U.K. Mobile than Fixed Broadband Users in 2011?

More people will use mobile broadband rather than fixed line broadband by 2011, mobileSquared predicts. It's the sort of shocking prediction that makes for a great headline, but also is misleading. The forecast, for the U.K. market, might lead one to conclude that users are disconnecting fixed broadband lines and using mobile instead.

But that is not what the forecast assumes. Rather, it primarily assumes continued growth of smartphone connections.

By 2011 the number of active 3G "smartphone" type devices in the UK will be 36.3 million. There also will be 6.4 million dongles and embedded devices in use, taking the total number of mobile broadband connections to 42.7 million compared to a base of fixed broadband connections of 42.5 million, mobile Squared projects.

To be sure, over time there will be more Internet access occuring from broadband-capable smartphones.
The firm estimates that between one percent and 10 percent of enterprise Internet traffic is already being generated from a mobile device, for example.

But most observers, and most users, likely would say that mobile broadband and fixed broadband are complementary, more than substitutes.

That noted, the application profile for mobile broadband likely will be distinctive. “Mobile will become the primary access point for brands and businesses communicating with its consumers within two years,” says Nick Lane, mobileSquared chief analyst. “Mobile is always-on, and the average user carries their device for an average of 16 hours a day. So if a company or brand is not already considering how to use mobile, then they need to because their customers are.”

As the typical mobile "phone" becomes a multi-purpose broadband device, it will be used for Internet applications. That is not to say the typical smartphone will replace a PC, or a fixed broadband connection. The application profile and mode of use will start to overlap. But each mode will retain key advantages for the bulk of usage. People will talk more on their mobile phones than on their PCs.

They will engage in research, document, calculation or process intensive operations, plus most long-form TV, on a PC or a notebook equipped for broadband access. But people will rely increasingly on their mobiles for social networking updates, location-related apps, real-time information and brief entertainment episodes, and sometimes for long-form video.

The point is that mobile broadband now consists of two distinct segments: smartphones and PC dongles. And while both overlap at times with fixed broadband, they are distinct. Wireless broadband used to connect PCs generally is a complement to fixed broadband access, not a substitute, though that will happen at times.

So the mobileSqured forecast, which essentially lumps all smartphone data accounts with PC dongle accounts to reach the conclusion that mobile broadband will be a bigger business than fixed broadband, is correct in one sense, but wrong in another. In fact, all forms of broadband access are increasing.

Order of Magnitude Increase in Mobile Bandwidth by 2015

U.S. mobile carrier traffic reach 724 TBytes per month in 2015, up from 15 TB per month in 2009, says Coda Research Consultancy, an overall compound annual growth rate of 90 percent. As you might expect, video is behind the sharp rise, growing 104 percent.

(click image for larger view)

In fact, by 2015, video is forecast to represent about two thirds of all traffic, or 459 TB a month, Coda Research says.

That could be a problem. Sprint and Clearwire, for example, argue their new WiMAX networks will provide three times to five times the bandwidth of 3G networks. That's all well and good, but this forecast suggests aggregate demand will grow 10 times from present levels.

Several issues: is there enough spectrum to handle all this growth? What is the cost of upgrading facilities, even if there is enough spectrum? Is there any video revenue model that pays for the investments? Will the mobile regulatory framework provide incentives or disincentives for investment?

Those are the big challenges. But the report has other nuggets for some parts of the mobile ecosystem.

Handsets will drive 68 percent of mobile carrier traffic by 2015, while netbooks will represent 14 percent.
Coda also estimates that mobile ad revenues will total $5.05 billion in 2015.

By 2015, mobile data revenue will grow to 47 percent of mobile service provider revenues, while voice generates 53 percent of revenue. This last prediction is important, as it suggests how and when mobile service providers will cope with the ultimate shift away from mobile voice business models.

How Not to Sell Hosted VoIP to Smaller Businesses

The single biggest mistake retail providers make when trying to sell hosted IP telephony to small and mid-sized businesses is that salespeople start with features, when they should start by reassuring buyers that “it is a reliable phone system,” Savatar VP Mike Ahearn told an audience of small telcos and cable companies attending a MetaSwitch marketing seminar.

The sales pitch has to begin with “it’s a high-quality phone system that is reliable and lets you keep your phone number,” says Chris Carabello, Meta Switch marketing director. In fact, establishing this lead proposition is so important retail sales personnel should establish that fact even before going to the “it will save you money” pitch.

Only after those two positioning efforts should sales personnel then add that hosted IP telephony “makes your life simpler.” The very last thing that should be discussed is that IP telephony offers new features.

And even then, when working with small business customers, even the discussion of new features should focus on a few new features that might appeal to the particular prospect.

That often is the reverse of the pitch made by many sales people, who lead with features first, says Carabello. The key message sometimes occurs at the very end of a discussion, but it needs to be delivered right up front, he adds.

It might seem unnecessary to emphasize that the product is “a managed, hosted IP telephony service that allows you to make and receive calls on IP phones or your computer,” but potential buyers are being asked to make a change in behavior that automatically raises the question of how well it will work.

And though there is greater understanding now that hosted IP telephony actually works, possibly 22 percent of potential buyers continue to think VoIP suffers from major quality of service issues, says Ahearn. As many as 28 percent to 32 percent of potential buyers with 100 or fewer employees might believe that, so take the issue head on, right away, he adds.

The important implication is that every prospect has to be reassured, right up front, that “it works.” Conversely, “cost savings” are generally seen as an IP telephony value.

And though hosted IP telephony obviously provides a path to selling many other services that could range from Web hosting to email and data services, “make the hosted telephony sale first, then up-sell later,” suggests Carabello.

The generic pitch should begin with the notion that hosted IP telephony “is an easy to manage phone system that will save money and help you run your business more efficiently,” says Ahearn.

Only after that is established should the salesperson move to the fact that it uses the Internet connection to make calls. And since most small businesses buy on the basis of a basic “cost per employee per month,” emphasize that hosted IP telephony offers a lower cost per user per month than the existing solution.

But there is one prevalent fact that suggests a simple SIP trunking offer will resonate with small businesses who already have invested in IP PBX gear. Ahearn points out that the trunk-to-phone ratio for smaller businesses is pretty close to 1:1. But an IP phone system really does quite well with a 4:1 concentration ratio.

Firms with four to seven employees report buying one to 1.2 trunk lines for every phone in use. Firms with eight to 10 employees report having 0.6 to 0.7 trunk lines for every phone in service.

Organizations with 11 to 20 employees report having 0.5 to 0.6 trunk lines for every phone. Firms with 20 or more employees say they have about 0.4 trunk lines for every phone.

The implications are fairly clear. Organizations that need to support between a few trunks and 14 PBX trunks are vastly over-provisioning trunk capacity. The typical organization using IP phones can get along fine with a 4:1 ratio of phones to trunks.

For a firm supporting six phones, and buying six trunks, an alternative SIP trunk strategy could save as much as $1,915 a year.

An organization requiring 14 trunks could save $2,205 a year by swapping SIP trunks with a 4:1 concentration ratio for PRI trunks that are provisioned at a 0.4 concentration ratio of phones to trunks.

The clear implication is that a small organization can save money immediately by replacing PBX trunks with SIP trunks.

Telco Business Models Diverge

Until recently, most global communications providers had business models that were highly similar. These days, it is clear enough that providers are starting to differentiate, and that the future business will feature several to many different business models.

Four telco business models will exist in the future, says Forrester Research analyst Mike Cansfield. Some carriers will stick with the vertically integrated model of the past, because they still can make it work. You will tend to see this most frequently among the largest global carriers, with the biggest customer bases and very large bases of recurring revenue.

Others will move to a partnership-based model, where some functions previously conducted in-house are shifted to business partners. Smaller national carriers with moderate customer bases will frequently use this model, as will carriers making aggressive expansion moves outside their historic footprints.

Some might shift to a horizontal model, though Cansfield points out that no legacy telco has actually decided to do so. This approach has been tried by some new competitors though. Vanco, which knit together a global VPN capability, is one example.

In the United Kingdom, other new contestants have chosen this approach, including Tesco, the supermarket chain, as well as the U.K. Post Office, says Cansfield.

In the mobility space the mobile virtual network operator model uses the horizontal approach. No major established operators have yet shifted from a vertically-integrated model to the horizontal model, though in some respects the “functional separation” model or “structural separation” model is an example.

The disaggregated model likewise is mostly a concept at the moment, not a practical option.

The horizontal model splits the network from the retail business, but in the future it will be easier to consider, if not adopt, a very-disaggregated approach where different functions are assembled on a virtualized basis.

This is a sort of cloud computing or “software as a service” concept applied in a very big way and perhaps can be thought of as the partnership model on steroids.

This fourth variant is based on the premise that a telco has a choice, says Cansfield. Does it own, operate, and manage a network within a horizontal structure or not? If it decides on the latter, then it can choose to disaggregate itself and find partners/outsourcers that can provide more or less all things.

The issue for the latter three models is that ownership of access assets remains valuable, and most would likely say strategic. Ubiquitous wired access networks are so expensive there always will be few of them.

Spectrum rights likewise are relatively scarce and expensive to acquire. Most executives probably would agree ownership of such assets, when possible, confers clear business advantage that should not be disaggregated. Some executives in some countries do not have a choice, of course.

The vertically-integrated communication service provider model is not going away, Cansfield argues, though it will not be the only model. Some providers will largely be able to retain the traditional model, where one entity controls the channel; owns, operates, and manages the technology deployed (usually meaning fixed and mobile); and runs the underlying network that delivers services.

The reason is simply that the networking business remains one where scale economies exist, allowing a large provider to operate efficiently where smaller providers simply cannot. A large provider of services to wholesale customers, enterprise, smaller business and consumers can leverage investments across multiple customer segments where a smaller provider cannot.

So the former incumbents of the world clearly will be prominent users of this model. Still, it is more than a semantic shift to note that the network becomes a platform in the new model, not the center of the model. Software, content and many new applications partially created by end users will be key.

That said, scale in and of itself will prove necessary but not sufficient. Carriers still will have to leverage scale to meet customer demand better than other providers can.

The multiple models also will lead to changes in performance metrics. While traditional financial performance is key for all contestants, there are changes in the need to measure product profitability and network performance, Cansfield argues.

This might sound odd, but what he appears to mean is not so much that the profitability of any single application or service need not be measured, but rather that it should be increasingly possible to gain visibility into the real costs and real profit margin for any service when providers gain the increased visibility many of the models allow.

That isn’t to say any service provider can dispense with a need to measure network and element performance.

At the network layer, measures like jitter and latency will clearly remain important. But Cansfield says other operating metrics assume new importance.

Non-traditional measures such as time between “lead to cash” also are measures of effectiveness. Likewise, the time taken from order to receipt of payment; cost to serve, or discrete analysis of how much it really costs to provide service to a specific customer and cost per transaction are better measurements of provider effectiveness.

This approach will enable the telco to benchmark itself against Google rather than other “me-too” operators, he says.

Those types of analysis are easier when a service provider actually sources inputs from partners, as there is a measurable and discrete cost. The traditional problem with conducting analysis at this level is that the traditional vertically-integrated model requires “guessing”: costs largely are allocations. And allocations inherently are political, based on any number of formulas that may not reflect the actual cost to create a product, sell and support it.

Getting a better handle on transaction or sales costs also is required, so service providers can derive unit total costs per service, a key step in understanding and then maintaining profitability, Cansfield says.

That is important if one assumes that retail pricing for products will decline over time. If that happens, more efficient sales and provisioning mechanisms are needed.

Rather than just focusing on metrics like financials, network performance and customer retention, new metrics also are needed. Measures such as cost per transaction, discrete customer profitability, and returns from bundles become important, he says.

The changes are propelled by choices in revenue dynamics. “Only five years ago, voice revenues at British Telecom amounted to 45.4 percent of total revenues,” says Cansfield. “In 2007 and 2008, BT voice revenues accounted for 39.3 percent of the total.” And virtually nobody thinks the basic trend can be reversed, though many think it will stabilize at some point.

But some other changes suggest where the communications industry already is headed. “Communications is no longer a discrete sector,” says Cansfield. That might overstate the case, but it is the direction things are moving as we move from single-purpose networks to multi-purpose networks.

83% of Enterprises Have Deployed Unified Communications

About 83 percent of 745 North American enterprise and mid-market executives have unified communications capabilities in place, or are planning to, while 17 percent report they still are not interested, says Henry Dewing, Forrester Research analyst.

Web conferencing and collaboration services, though, are seen as a priority by 55 percent of SMB respondents, as well as storage and backup services, also seen as a priority by 55 percent of SMB respondents.

Integrated communications that unify voice, email and instant messaging are the most-wanted capabilities, with twice the number of executives saying that is important, compared to other features such as presence, integration with business applications, can conferencing capabilities.

That isn’t to say there is little or no interest in features such as desktop call control or mobile integration, but that demand for those features is about 2.5 times less important than unified handling of voice, email and IM traffic.

And while demand for specific features is relatively unevenly distributed, the business value drivers are fairly broadly distributed. Saving money, providing better customer service, improving communication flows and saving time all are cited as key values.

At a time of very-tight information technology budgets, more than a third of respondents say they are hiking spending on hardware, servers and desktop software. About 15 percent report they are increasing spending for managed UC services.

The situation at small and medium-sized businesses and organizations is a bit different, as you might suspect. Where 83 percent of enterprises have unified communications projects in place or in progress, only about 24 percent of SMBs say that is the case at their organizations.

And about 20 percent of SMB executives surveyed say they really have no interest in UC.

And though it seems logical to many of us that SMBs remain prime candidates for hosted services that avoid major capital investments, most SMB executives say they are more interested in premises-based solutions.

When asked how interested they are in buying a managed UC solution sometime in the next 12 months, 56 percent of SMB executives say they “are not interested.”

About 21 percent say they are “somewhat” interested while four percent say they are “very interested.” About 11 percent of SMB executives surveyed by Forrester Research say they currently are using a hosted UC solution.

So it appears industry advocates have some ways yet to go in convincing SMB executives that hosted UC solutions are a better approach than premises solutions.

Recent surveys of IP telephony adoption by SMBs have suggested a similar attitude towards hosted IP telephony as well. About a quarter of SMB executives say they would consider a hosted IP telephony solution, while about three quarters still say they would be more comfortable with a premises-based solution.

Call it habit, inertia or lack of trust. SMB executives still have not embraced hosted IP telephony at rates many of us expected. Some have suggested that fear about making a mistake with a mission critical tool is compounded by fear of choosing the wrong supplier.

Extreme fragmentation of the supplier base, as also is typical of the information technology support business, means no single name generally stands out—in the service provider space—as a “logical” supplier of IP telephony or unified communications.

On the other hand, buyers seem more familiar with the brand names of the firms supplying them phone systems, which then are likely vehicles for a move to IP telephony or unified communications as well.

So far, the hosted IP telephony industry does not seem to have tipped the scales, though one might argue that 25 percent penetration of the customer base for a relatively new solution is not shabby.

Is Rural Broadband Penetration Close to 100 Percent?

Is it possible that rural broadband penetration actually is pretty close to the penetration of Internet users? In other words, is it possible that use of broadband in rural areas now is close to 100 percent of Internet users?

New data from comScore suggests that might be closer to the truth than many believe. The latest estimates are that, in rural areas, broadband penetration is at 75 percent. If one assumes some rural users still use dial-up, that suggests perhaps 85 percent of rural households now use the Internet.

In 2007 the U.S. Department of Agriculture Economic Research Service estimated that 63 percent of all rural households had at least one member access the Internet.

If rural broadband penetration now is up to 75 percent, as comScore indicates, that would imply that Internet usage is at least that high, in other words.

That would seem to have implications both for setting of national broadband policy and policy in rural areas. For starters, the new data suggest that rural broadband is growing robustly, without any additional government activity.

Some might argue that broadband usage remains lower in rural areas than in metro areas, and that remains true. Metro broadband penetration is at 89 percent. But virtually every study has shown that Internet usage also is lower in rural areas. The point? Lower Internet usage obviously means lower broadband access penetration.

One has to be careful with statistics, though. By definition, a household with no ability to access the Internet would not be an Internet-using household. So a better way to describe comScore’s findings are that, when wired facilities are available, rural households are buying broadband at rates not dissimilar to urban users.
That isn’t to say adoption is equal to urban rates, but that the gap is closing awfully fast.

Broadband penetration in U.S. rural areas increased 16 percent from 2007 to 2009, while metro area broadband penetration grew 11 percent, according to comScore.

In part, that is because rural markets have more room to grow. The analogy is wireless voice growth, which is highest in places such as India, China and Africa, where penetration is lowest.

“With low-speed DSL priced at about the same level as dial-up in many areas, there is little incentive for households to remain on dial-up,” says Brian Urutka, comScore VP.

Rural markets with a population less than 10,000 grew broadband penetration by 16 percentage points. Areas with population between 10,000-50,000 grew penetration 14 percentage points while metropolitan areas with populations of 50,000 or more grew penetration by 11 percentage points.

Critics sometimes say that even if access is not a problem, access speeds are, and that is an argument that makes sense. The issue there, though, quite often is the “middle mile” trunking between major points of presence and the actual rural communities.

Basically, the problem is not the Internet backbones, and not even so much the local access networks, as it is the trunking network to backhaul traffic to the Internet PoPs. Many rural ISPs find, for example, that they have access to a T1 or two T1s in the middle mile. That makes it tough to deliver faster broadband access to customers on the local access networks, for obvious reasons.

The Internet backbone is a firehouse. So are the access networks, for the most part. But the middle mile is a straw.

Solve the middle-mile problem and broadband access probably ceases to be an issue in many communities. Yonder Media CEO Craig Vallarino estimates that half the cost of building fixed wireless networks in rural areas is in the core network and middle mile.

The radio infrastructure represents about 20 percent of cost while customer premises investment represents about 30 percent of cost. In other words, it isn’t the access infrastructure which is the main investment barrier: the middle mile is.

That said, there still will be some locations so isolated that only a satellite connection really will ever make sense.