We use cookies and other similar technologies (Cookies) to enhance your experience and to provide you with relevant content and ads. By using our website, you are agreeing to the use of Cookies. You can change your settings at any time. Cookie Policy.

The Connected Web

Phil Wainewright

2010 Watchlist: Mining the Off-Peak Cloud

Vote 0 Votes

Over the next week or so, I'm going to highlight several emerging trends in computing that I believe will be very much worth watching in the coming year as important sources of innovation and competitive advantage for those who harness them.

The first on my list is the most recent to emerge — even though it's been something I've been expecting for all of the last decade, ever since it first became obvious that Internet technologies would make it possible to access computing resources from anywhere in the globally connected Web. It was always self-evident that some of those resources would have spare capacity at certain times of the day, and that we would start to see providers offering off-peak pricing to encourage take-up of that spare capacity for others to mine.

This month, Amazon fulfilled those expectations with its announcement of spot pricing for its EC2 cloud computing service. Under the new arrangement, Amazon will make spare capacity available at automatically determined 'spot' pricing, and will sell it to customers who have bid the same price or higher. As several observers have pointed out, this should be a lucrative trade for Amazon, which is selling off capacity that would otherwise be idle for the best price it can get. It's far from commoditization. But as Dion Hinchcliffe notes in his ebizQ commentary on the announcement, "you can count on ... competitors to respond, which is where it gets a lot more interesting ... it won't take long for customers to catch-on that real value is to be had when pricing is tied more closely to what cloud resources are actually worth at any given time."

I suspect that developers building on-demand services will be among the first to catch on. Enterprises will find it harder to rearchitect their existing applications, but anyone developing a new application will be able to build it into their design from the outset. This opens up some interesting new opportunities for developers that can leverage the lower prices available when using off-peak compute capacity.

It will especially make sense for applications where the processing of data doesn't have to take place at the same time that it's collected. Payroll, incentive payments and expense claims, for example, are common transactions that are traditionally 'batched up' for processing. Daily, weekly or monthly activity reports can similarly afford to wait their turn for at least a few hours. There are probably much bigger examples in other fields. I can imagine that refreshing template items across a large website might be an action that would cost much less using spot pricing than it would on-demand, especially if this were a frequently recurring action. (If you're thinking of other compelling examples, please add them in the comments below).

Off-peak cloud offerings from other providers than Amazon may add more tweaks to the model, such as a lower price for a less rigorous service level agreement (SLA). After all, if you're willing to wait for batch processing, in many cases you'll also be willing to wait an extra hour or two through unexpected downtime (and most times you won't even be affected by it).

Whatever permutation you choose, the ability to mine sources of off-peak computing opens up the opportunity to offer services with a much lower cost of operation, thus either undermining competition with lower pricing or simply improving profitability (perhaps both at the same time). Developers who do this will form a growing market for off-peak cloud computing that will act as a powerful magnet for further commoditization.

We are still some way distant, though, from a time when it will be easy to switch from one cloud computing provider to another. There is a lack of standards, not only in technology terms, but also in calibrating service levels and pricing metrics between providers. Three years ago, ComputerWorld's Mark Hall wrote:

"For utility computing to fulfill its promise, there has to be a standard pricing model that all users can apply to their operations. Until then, on-demand computing will be just another complex, proprietary pricing strategy vendors use to keep you from fairly and accurately comparing one service to another."

We are no closer to a standard pricing model for cloud computing today than we were then, and with no easy way to compare services, there's no easy way to automate switching from one to another. This provides an interesting opportunity for intermediaries, such as startup CloudKick, whose cloud management service lets you Hop to a Cheaper Cloud Today, as I described it earlier this year. These and other services (such as Conformity, whose founder recently recorded a podcast here) will do the groundwork and standards will eventually emerge. We won't see the fruits of that in 2010, but we will see early steps towards it, with the first examples coming forward of SaaS providers using off-peak cloud computing for competitive edge.

Also on the 2010 Watchlist: Ubiquitous Collaboration and People-Oriented Architecture.


Spot pricing is of course an interesting development, but maybe it is not economically very important. The biggest costs of IT are humans - programmers and system administrators and software licenses. Therefore saving some (tens of) percents on hardware might save you only few percents on the overall cost of the system. If you take into account that for saving with the spot pricing the overall system must be made more complex then there might even be no savings at all - the money you save on hardware must be spent on additional programming, installation and configuration.

Test/dev is another application area that could take advantage of off-peak pricing, at least for larger runs that are often timed to run overnight anyway (eg, HPC, some types of financial model development, etc).


Phil Wainewright blogs about how businesses are using the Web to get better plugged into today's fast-moving, digital economy.

Phil Wainewright

Phil Wainewright specializes in on-demand services View more

Recently Commented On

Recent Webinars


    Monthly Archives