Net Neutrality and the Cloud

As we stated in our previous post about WikiLeaks, we do not do politics here at The Virtualization Practice. But we do analyze the technical and economic impacts of issues which may appear in the political realm. With this as our charter, we take a look at the current state of the Net Neutrality debate and decisions and their potential impact upon the evolution of Cloud Computing.

Framing the Conversation

First of let’s try to frame up the conversation with a set of facts and hopefully non-controversial assertions:

  1. Cloud Computing will ride on the same Internet that we know and use today – or at least a version of it that has evolved in the future from what we have today
  2. Therefore Cloud Computing will be viewed by its consumers as “just another Internet application” like web browsing, email, VOIP, instant messaging, social media (Facebook, Twitter, etc.) and various forms of Internet video
  3. Most consumers of the Internet (be they individuals, households, small businesses or large enterprises) pay flat rates for Internet access per month based upon the capacity of their Internet connection. This is true for households who buy cable or DSL access, enterprises who buy speeds from T-1 on up, and mobile smartphone users who typically pay $30 a month for a unlimited data plan.
  4. Although most consumers of the Internet pay flat monthly rates for “all you can eat” plans, there are some restrictions that go along with “all you can eat”. For example most ISP’s do not allow people on consumer plans to host web servers in their homes, reasoning that one media server in one house could consume a great deal of the bandwidth allocated to an entire neighborhood. Google Apps for Business users can only send email to 2000 different recipients in a day – with the obvious rationale being that more than this starts to look like a commercial newsletter service for which one should pay more than $50 a year.
  5. Providers of content on the Internet do not pay for the Internet in any kind of proportion to the amount of the Internet’s capacity that they use. If you host a web site you “pay for bandwidth” but what you are paying for is a portion of what it costs your ISP to run fat pipes into their data center. If all you have is a blog what you pay may be no more than a few dollars a month. If you are YouTube, NetFlix, or Hulu you probably pay for hosting plus the bandwidth into your servers, but you do not pay for the bandwidth that you consume from  your servers all of the way to the consumer.
  6. Providing increases in bandwidth all of the way to the end users of the Internet is expensive. For example, Verizon is reported to have spent $23 billion building out its FIOS service which amounted to $4,000 for every potential consumer. Of course these numbers do not include the degree to which the backbone of the Internet that serves Verizon would have to be upgraded as substantial numbers of Verizon’s customers start to use all that available last mile bandwidth.
  7. According to Cisco’s Virtual Networking Index, mobile data traffic will double every year through 2014 reaching 3.6 exabytes (3.6 x 1018) per month by 2014, and almost 66% of the mobile data traffic by 2014 will be video.
  8. A ubiquitous and fast Internet is an essential productivity tool for the way in which many people in the USA (and in many other developed countries in the world) work today. Working out of our homes and wherever we happen to be at all hours of the day has now become the accepted norm for many people and many industries.

The State of Bandwidth Today

Given that bandwidth is important and that more if it appears to be essential as we continue to develop new uses for the Internet where do we stand today? The Organization for Economic Cooperation and Development (OECD) has done the following analysis of average broadband speeds for 11 major global economies.

Worldwide Average Broadband Speeds

If you live in the United States (where the Net Neutrality debate and proposed regulations are currently occurring), the above chart is not good news. The USA is in 9th place among 11 global economies in terms of bandwidth available to the average consumer. Part of this is due to the fact that with a large area to cover and a distributed population, it will be very expensive to run fiber to every household in the US. Part of it is also due to policies on the part of past administrations that had the economic effect of discouraging investment in our Internet backbone.

Who and What will Drive Bandwidth Consumption?

Cisco’s Virtual Networking Index predicts that in global Internet usage, Internet Video in a variety of forms is going to comprise almost half of bandwidth usage in 2011, and almost 60% of bandwidth usage in 2014. The numbers below are in Petabytes per month where one Petabyte is equal to 1000 Terrabytes or 1,000,000,000,000,000 (1015) bytes.

Global Consumer Internet Traffic PB per Month2

Notice that the three fastest growing categories of bandwidth consumption are all related to video of some kind. If we graph this the picture below emerges. The three categories of video are all in various shades of blue and are the middle of the graph. Video is growing so rapidly that all other categories are shrinking as a proportion of the total over time.

Global Consumer Internet Traffic Growth

Back to Net Neutrality and the Cloud

If you accept the assertions that 1) demand for Internet bandwidth is going to grow (at an overall rate of 36% per year for the next four years) – driven primarily by video, 2) Cloud Computing will be an Internet application that will rely upon a fast and ubiquitous Internet to work, 3) that the USA is not leading the world in bandwidth available to its consumers today, and 4) investing in improving Internet speeds is expensive then this all then reduces down to how do we best ensure that the Internet is able to provide an acceptable transport for the delivery of those next generation cloud services that we are all counting on consuming and producing?

The above question about ensuring that the Internet grows to meet the needs of Cloud Computing really then reduces down to two sub-questions:

  • What is the best mechanism to ensure that the investment occurs that is necessary in order for the Internet to grow its capacity in a sufficient manner?
  • How should the question of who pays for what part of the build out of the Internet get resolved? This question itself is broken into two parts, 1) who pays, and 2) who makes the decision as to who pays for what and how much?

Ensuring Adequate Investment in the Internet

In a free market system, the process of deciding whether or not something gets more or less investment is pretty simple to describe. Investments that promise more profits to investors get more investment capital than investments that promise less or no profits to investors. Given that there is no shortage of investment capital in the world (US corporations are currently holding approximately $2T of funds in cash or near cash vehicles waiting for the right opportunities and investment climate to materialize), it would seem that one approach would be to simply let the free market price Internet access (to both consumers and producers of content) in a manner that ensures that the required profit streams exist.

However, it is in the question of who decides how much both consumers and producers are going to pay in the future that the question gets tricky. This not just a simple question of whether or not Comcast will be allowed to charge more if they provide an Internet service that has twice or four times the bandwidth of the current service. This is also a question of whether different content providers will have to pay based upon how much bandwidth their content uses. This question arises because some content (video) is going to consume so much more of the bandwidth than everything else, that it is only logical to ask if video providers are going to help pay for the pipes to carry their movies to their consumers.

The Free Rider Problem

In economics, there is something known as a free rider problem. The free rider problem comes into being whenever a good or service is made available either to everyone (Interstate Highways) or to large numbers of people with flat rate pricing (broadband Internet access) and it is possible for some people to consume vastly more of the broadly available good than most of the people do.  In the case of the USA highway system, it was determined that large trucks were causing wear and tear on the highways to a far greater degree than automobiles. Therefore provisions were made to tax truck operators so that funds could be collected from them that helped pay for the extra costs that their trucks imposed upon maintaining the highway system.

In the case of the Internet, video is our free rider. Netflix and YouTube pay nothing to the Internet backbone operators or the providers of Internet access to consumers like Comcast, ATT, and Verizon for the privilege of running their content through the same pipes that carry email and the rest of the web traffic despite the fact that video is already consuming half of the capacity of the Internet and is projected to consume close to 60% of that capacity in 2014. Substantial investments in the Internet will need to be made in order for it to carry the video content that is projected to be available and demanded by consumers – and the Internet will have to grow in capacity to be able to carry this content or else there will not be the bandwidth available to carry the cloud based services (some of which may include video) that we are counting on.

So Who Should Decide?

There are obviously two choices that we as an industry and we as a nation face when it comes to this decision. The government has been in and out of this process depending upon what flavor of regulatory regime has been in power in Washington. During the Clinton years, the FCC forced managed competition upon the phone companies and forced them to lease their data lines at wholesale costs to competitive local phone companies (remember CLEC’s?). This reduced the profitability of data services to the phone companies and resulted in them investing less in new network capacity. Many view this failed policy as the reason we are no longer a leader in bandwidth.

After the “managed competition” debacle the approach was very much hands off. One could argue that for the most part investment in the capacity of the Internet backbone has kept pace with the demands placed upon the Internet in the last 10 years. However, the principle concerns on the part of net neutrality advocates were not really about capacity, but rather that the process of potentially limiting the bandwidth of a class of user (video providers) was analogous to limiting bandwidth to a “point of view”  which turned this into a freedom of speech issue that should be made on a political and not purely economic or financial basis.

With all of this as the background we now face the fact that the current FCC has interjected itself into this decision. The FCC has decided that it will enforce some form of net neutrality. What this means in practice is hard to predict – but the basic principle is that the FCC will via regulation force network providers to treat all content as being the same except when the FCC decides to make an exception. Whether this is a good thing or a bad thing probably depends upon your view of free markets and unintended consequences of regulation and is outside of the scope of this article. However, regulation of the Internet will have some predictable impacts upon the Internet and therefore upon the evolution of Cloud Computing.

Summary Impact Upon Cloud Computing

If the FCC proceeds down a true net neutrality path, then this will clearly have a set of impacts upon the state of the Internet as a platform for Cloud Computing. Those impacts are most likely to be:

  • In general whenever the government regulates something the free market provides less of it. It is hard to find an example of where investment in something has increased due to government regulation (the government can increase the supply of something by subsidizing it as it has done with ethanol – but there is no discussion of subsidizing the construction of the Internet as a part of the net neutrality debate).
  • The growth in video content will demand dramatic increases in Internet bandwidth in the very near future.
  • Cloud Computing in its most basic form (the access to an application over the Internet) will rely upon an Internet that is fast enough on a consistent enough basis so that people are willing to trust it as the transport for their online business activities
  • Advanced forms of Cloud Computing may well incorporate video services – putting Cloud Computing in dual role of being a victim of video’s appetite for bandwidth and a contributor to the problem
  • It would seem that providers of premium video services like Netflix should have an incentive to ensure an adequate transport for their offerings to their customers and ought to be given a mechanism to help fund the expansion of the Internet in a manner suitable to their needs.
  • It would seem that network operators (both backbone and access providers) ought to be allowed to charge the “free rider” (video) fees that are proportional to the degree that video consumes large portions of the bandwidth in these networks.
  • If network operators are not able to generate a return on the investment in incremental capacity from the largest users of that capacity (video providers), then this may hurt the ROI on these investments in the same way that managed competition did the last time the FCC tried to regulate the Internet. The problem is at least twice as complex as it was last time the FCC tried to regulate the Internet since now both fixed land line and mobile Internet capacity needs to grow at rapid rates in order to keep up with the projected demand.
  • If the ROI on these investments is hurt by FCC regulation then this could have the effect of diminishing the growth of Internet capacity at the exact time that both video services and emerging Cloud Computing services are demanding dramatic increases in that capacity.


Cloud Computing will rely upon adequate Internet bandwidth being available to ensure that users of cloud computing services have an acceptable end user experience, and that providers of these services can promise that acceptable experience to their customers. The Internet (both the wired and the wireless parts) needs a funding model that ensures that the right capacity is available to the applications (cloud services) that need it. We face a choice as to whether or not FCC regulation will help or hurt the progress towards the correct funding (and ultimately payment for Internet services model). Choosing incorrectly may severely inhibit the growth of Cloud Computing services due to confusion over bandwidth and pricing decisions. In particular having the government declare all forms of content to be equal may stop network providers from being to charge certain types of content (video) for their actual share of the bandwidth – diminishing the ROI on investments in bandwidth – which would in turn harm the creation of the very bandwidth that the forthcoming Cloud Computing industry will have to rely upon to flourish.

Posted in SDDC & Hybrid CloudTagged , , ,