There are myriad ways to bill colocation customers, making a comparison between multiple bids an occasionally daunting process. The industry does seem to be shifting towards an accepted standard billing model based on metered electricity use, but older billing methods based on footprint and telecom connections are still in play. What are the distinctions in colocation pricing models and what is the fairest method for customers and providers alike?
Virtually all colocation bids will be based on one or more of the following: (1) amount of floor, cage, or pod space used; (2) cross-connects or the amount of bandwidth; (3) energy consumed; and (4) labor. Historically, colocation started off with bandwidth at a high premium. Fifteen years ago the cost of a 1 Mbps connection was significantly larger, so it made sense to bill by the amount of information transferred over the network. Once bandwidth became cheap, servers and equipment were still bulky and inefficient. Operators starting billing based on the amount of real estate used by a customer—the more racks, the more expensive.
A modern data center is typically outfitted with blade servers in high-density arrangements, meaning the amount of space needed for significant power isn’t as much as it used to be. To compensate, data center providers would often charge a minimum kW per cabinet or rack, making up for power density. More and more operators are switching to the (arguably) more fair and measurable approach: charging by power use.
There are two main ways to charge for power usage: per-whip and metered. Charging per-whip is just a flat fee each month to supply power to equipment, while metered power charges for each kW consumed. There are frequently minimum monthly charges included in a metered billing model.
Gartner recommends three colocation pricing models depending on the size of deployment, but all of them boil down to basically the same formula: A space fee + (meter reading for IT equipment * PUE). The space fee covers security, building operations, etc, and is based on the power draw estimate provided by the customer and the provisioning density as estimated by the provider. (There’s no escaping the real estate fees completely, after all.) The PUE multiplication is included to cover infrastructure use beyond IT equipment, like cooling.
Request more information from the experts.
One final pricing model is “all-inclusive”, including managed services, fuel for generators, taxes, installation—basically factoring in the customer deployment as a percentage of the overall data center operations.
Each of these aspects can also be added to any of the previously mentioned billing models, including pricing by energy meter. Many metered contracts will also include managed services, either picked by the customer or as a block of pre-billed hours for hands-on technical support.
Installation and setup charges are separate from monthly recurring cost, but should be considered as well. Meanwhile, providers in Europe and the United States often charge differently for different items. One major variation is cross-connect charges (which allow customers to use more than one telecom provider). Recently, European providers have been more likely to charge higher prices for energy or space, while American providers hike up the price of cross-connects. In the end, this comes out largely a wash.
The pricing model for colocation has largely shifted along with trends in the industry, mostly dependent on technological limitations. First bandwidth was the bottleneck and cost driver, then server space. Now power is the premium. Perhaps one day we’ll see a clean-energy future, where power is abundant. What do you think will be the basis of colocation billing should that day come to pass?
Posted By: Joe Kozlowicz