Bandwidth Management, Cloud & the 405 Freeway

Bob Deutsche joined Intel in 2004 and has more than 25 years of business and IT experience in positions that ranged from data center operations to software development to CIO. He can be found online at Bob Deutsche on the Intel Server Room.

Bob_Deutsche_sm

Like so many people around the world, you may have visited Los Angeles at some point - if only for the weather, especially this time of year. Whether your visit was for business or pleasure, you likely drove on what we Californians call The 405 (also known as the San Diego Freeway). Separating greater Los Angeles from beach communities like Malibu, Santa Monica and Redondo Beach, The 405 is among the top five most congested freeways in the U.S. With its last upgrade completed in 1969, The 405 just wasn’t designed to accommodate today’s traffic flow. While you can make forward progress (assuming no accidents or obstacles are blocking any of the 12 traffic lanes), you’ll likely go at a pace that is…well, let’s just call it "leisurely."

This may not be a perfect analogy, but navigating through LA on The 405 seems an apt way to start discussing our seventh fundamental truth of cloud computing: Bandwidth and data transmission may not always be as inexpensive and unencumbered as they are today.

As discussed in my last column on cloud policy, bandwidth management is one of six government-regulated policy and standards considerations that you (and your cloud provider) must understand as you continue your journey to the cloud. Since every government has its own approach, the best advice I can offer is to balance your goals against your government’s related policies. One size definitely does not fit all.

Broadband Spectrum Considerations

To begin, it’s key to recognize that the broadband spectrum is a shared resource.

As a consumer, I expect that everyone in my household, on any number of devices, can simultaneously download a live video stream of our favorite movies or TV series, have instant audio and video chats, and keep track of our stock portfolios—all while playing video games. I also expect latency of less than 100 milliseconds and costs that I consider “fair,” with anything more being an example of the provider getting rich at my expense.

As a business, I understand cloud computing requires data center broadband connectivity to any number and/or type of devices, including those that can only receive data (zero/thin client). Further, these ecosystems must provide service and reliability that meets the specific needs of my business. For example, healthcare and command and control systems demand reliability and low latency. Considering my quality of service and reliability needs, I expect to pay accordingly but need the provider to disclose meaningful data such as service terms, pricing (perhaps based on volume rather than time), packet loss, delay, service availability rate, and SLA based response time (latency).

Evolving Revenue Models

With all of this in mind, it’s likely the revenue model for broadband services (which vary greatly from geography to geography) will soon change. At a Telco 2.0 conference earlier this year, I saw this chart:

telco-chart

The image suggests that by 2016, there will be around 15 times more mobile traffic than there is today, primarily driven by video (this is the traffic on the access side). More recent video traffic projections I’ve seen at a December 2011 Gartner Data Center Conference suggest this number may be conservative.

Historically, switched voice has been the cash cow for telecommunication companies. But as the image suggests, the projected growth in this traffic is flat. The company presenting the slide stressed that today’s broadband fees are assessed based on time, instead of network volume used.

Huge Amounts of Data Are Flowing Over Networks

As the conference attendees discussed this slide, an implied consensus emerged: This model will likely change because of the burgeoning amount of data transmitted over the networks (both wired and wireless).

There is some evidence that these changes are already happening, at least in the U.S. In May 2011, one of the large U.S. carriers began imposing monthly data limits for its fixed broadband subscribers. With U.S. consumers seeing broadband as described above, the move was highly unpopular. Again, it’s perhaps safe to conclude that tiered pricing packages of some sort—based on how much data is consumed or how fast the data is delivered, or a possible hybrid of both—will evolve over the next few years. Can I state this with absolute assurance? No. But then again, if there is a chance of it happening, how do I mitigate the risk to my company—particularly if I’m justifying the cloud based on ROI?

Another topic of some importance in the U.S. (primarily because as I understand, it has no jurisdiction over wireless providers) involves Net Neutrality. Net Neutrality is fundamentally about data capping. With an admission that this discussion is best left to attorneys, it’s still safe to conclude that you and your cloud service provider need to at least be aware of what’s happening under this umbrella.

To me, it seems prudent to anchor any expansion of business activity in your cloud—beginning in your data center and extending to how you design your cloud-based application portfolio and ending with the selection of intelligent end user devices—to at least a tacit understanding of how these topics may or may not impact your costs and success.

Finally, to help keep things in context, here’s the complete list of our inviolable, fundamental truths of corporate cloud strategy:

In future discussions, I’ll continue to detail these fundamental truths. As always, I’m interested in what you see or hear about today’s topic or others we’ve discussed. Please join in the discussion by posting a comment or contacting me via LinkedIn.