Amazon Sells Customers On Long-Term Use

As more Amazon customers sign up for routine use of cloud services, Cloudyn introduces a calculator to help choose cost-efficient mix of long-term and on-demand servers.

Charles Babcock, Editor at Large, Cloud

October 22, 2012

5 Min Read

Cloud computing customers tend to view it as a vehicle for absorbing their websites' or other public-facing applications' heavy traffic periods. But Amazon appears to be increasingly successful at getting more of them to use it for long-term, steady-state purposes.

The data is skimpy, and Amazon Web Services wouldn't divulge how much of its business is now based on Reserved Instance versus on-demand servers. On-demand servers are ordered up without prior notice; they start and stop whenever the customer wants. Reserved Instances, available since 2009, are lined up through one- or three-year agreements in exchange for an upfront payment.

Reserved Instances don't have to run continuously for that length of time, although one of the options for purchasing them, heavy utilization, assumes that they run most of the time.

The enticement: customers who use Reserved Instances can routinely save a third of the hourly cost of a similar sized, on-demand instance. Skillful use of Reserved Instances "can save customers up to 71% over on-demand rates. We first launched Reserved Instances based on requests from customers who had predictable, steady-state workloads and wanted to reserve capacity in exchange for lower costs," Amazon spokeswoman Tera Randall said in an email Oct. 19.

Last December, AWS then tailored Reserved Instance types to try to give customers better matches to the types of applications they were running. They could choose heavy-, medium-, or light-usage Reserved Instances, depending on the amount of time in a given day the server would run.

[ Want to learn more about how to calculate costs in the cloud? See Cloud's Thorniest Question: Does It Pay Off? ]

The additional choices appear to be encouraging use of Reserved Instances. Cloudyn, a cloud-use monitoring and analytics firm, said it has seen its customers' use of the service climb at a steep rate this year. At the start of the year, 28% of its customers were users of Reserved Instances. By the end of September, that figure had risen to 48%, said Sharon Wagner, CEO of the Israeli firm with 350 customers using AWS' EC2.

Cloudyn launched a Reserved Instance calculator Monday, a tool that makes use of an AWS customer's read-only credentials to import data on a customer's use of EC2 and recommend what it thinks are the most efficient Reserved Instances configurations.

Wagner said his firm's customers typically are spending $250,000 a year for cloud services. Use of Reserved Instances typically saves the average customer 30%, in exchange for an upfront payment of about 30% of the total bill, he said. Given the fact that Amazon phased in heavy-, medium-, and light utilization versions of Reserved Instances last December, the savings can be higher. The designations are measures of how many hours of the day a reserved server will run. Selecting the right type of Reserved Instance can seem complicated. Experience in using the cloud helps. But the tool offers the aid of a rules engine and advanced analytics in making decisions, Wagner said.

"Customers are not always selecting the right pricing model," from Cloudyn's point of view, he said. Cloudyn did an analysis of its 200 customers that it felt were most representative of the Amazon customer base and found 65% of Reserved Instance contracts were for one-year contracts; 35% for three years. Customers in many cases had a mix of both, with the ratio remaining roughly two to one in favor of one-year agreements. The greater potential savings lie with three-year contracts, Wagner said.

Customers might think signing up for a Reserved Instance and using it only 37% of the time would be a waste. But if they choose the light utilization option under Reserved Instances, they'd save 24% over relying on on-demand servers, indicates a chart showing Cloudyn efficient use cases. The calculation is based on using a large Linux server for a year in U.S. East 1, Amazon's Northern Virginia data center complex.

The Cloudyn monitoring service collects read-only data on all of a customer's servers flowing from Amazon's CloudWatch service. Cloudyn maps the server types and collects usage data, then applies a rules engine to the results. In many cases, the customer would be better served by Reserved Instances than on-demand servers, said Wagner.

"We think an Amazon user should target 50%-60% of servers as Reserved Instances--that would be the right balance," with the rest on-demand servers, he said. The higher the percentage of Reserved Instances a customer opts for, the higher may be the savings, provided he actually uses the contracted servers throughout the period signed up for.

In addition, many customers also choose the seemingly safe, middle-of-the-road medium utilization grade for their Reserved Instances when they would be operating more efficiently if they choose a server that was either light or heavy. Wagner said his customers appear to be clarifying how they plan to use Reserved Instances. Since May, use of medium instances has dropped from 35% to 19%. Light instance use has risen from 2% to 10%, while heavy instance use has risen from 26% to 36%, Wagner said.

Another cloud usage analysis firm, Newvem, has seen a similar uptick in the use of Reserved Instances. As customers get more experience in Amazon's cloud, they are beginning to evaluate alternatives to what they're doing to drive down costs, said Zev Laderman, CEO.

On Sept. 12, AWS launched a Reserved Instances Marketplace where one customer could sell the unused time remaining in his one- or three-year contract to another customer, if he had decided that he no longer needed the RI.

That takes away some of the risk of signing up for Reserved Instances, Laderman pointed out. It allows "no-regret purchases of Reserved Instances," where the customer saves over on-demand servers, even if they're not needed for the length of the contract. Customers can't always resell them for the full value of their remaining time in the marketplace, but they can still realize some payback on their investment.

"It's another way Amazon ensures the customer stays with them for the long haul," he added.

About the Author(s)

Charles Babcock

Editor at Large, Cloud

Charles Babcock is an editor-at-large for InformationWeek and author of Management Strategies for the Cloud Revolution, a McGraw-Hill book. He is the former editor-in-chief of Digital News, former software editor of Computerworld and former technology editor of Interactive Week. He is a graduate of Syracuse University where he obtained a bachelor's degree in journalism. He joined the publication in 2003.

Never Miss a Beat: Get a snapshot of the issues affecting the IT industry straight to your inbox.

You May Also Like


More Insights