Amazon Web Services prides itself on offering highly competitive prices to businesses of all sizes, but sometimes, customers engaging in the cloud migration process suffer sticker shock when they get their first bill after the migration has completed. The reality is, even the biggest AWS customers aren't exempt from making mistakes that can lead to disappointment and buyer's remorse when they do not realize the savings they expected. In fact, part of the current backlash regarding enterprise cloud migration is directly related to poor management of spending. When it's incredibly easy to access more resources as needed, it's also easy to let costs spin out of control, but fortunately, there are strategies for making the most of cloud services without missing the opportunity to enjoy substantial savings.
Many ways to pay for compute resources
To make the most out of AWS, organizations must understand the pricing structure. Customers who want to consume the elastic cloud by the hour with no commitment are wise to choose the on-demand pricing option. It's the selection most often used to fulfill unexpected, seasonal, or short term needs. It's probably also the most well-known offering since it fits the flexible, consumption-based billing model so widely advertised by Amazon. But it's far from the cheapest solution.
AWS Account Manager Boyd McGeachie strongly encourages enterprises to learn how reserved instances can fit with their business model to achieve greater savings. "In talking with clients, two things have become very obvious. First, while customers are very happy with the savings that they achieved by moving to Amazon, everyone always wants to spend less. The second thing is that reserved instances are highly misunderstood in the market. People aren't maximizing the dollar value of their AWS services." For a small financial commitment up front, companies can reserve instances for several years at a 50-75% savings over the cost of on-demand resources. This approach is ideal for steady state or recurring workloads.
Playing the market brings steep savings with little risk
Perhaps the most interesting and innovative option is the spot market. Since AWS is always on, the cloud provider often has unused capacity that can be obtained for bargain-basement prices. As Amazon Web Services (AWS) Senior Vice President Andy Jassy points out, "For those that have opportunistic workloads that can afford to be interrupted and where they want the lowest possible price, they often use our spot instances. We take all of our excess capacity available at any one moment and make it available on the spot market. Customers bid a maximum price they're willing to pay to run that hour. If they're over the market clearing price, they get to run. If they're not, they don't." Spot instances are often used for tasks that aren't time sensitive such as ongoing analytics for older data. Some customers even use these instances in development to spin up testing environments during times when overall demand on the compute market is low.
We take all of our excess capacity available at any one moment and make it available on the spot market.
AWS Senior VP
Keeping up with the pace of change
One common reason enterprises pay too much for cloud services is lack of attention to the changes going on at Amazon. AWS prices decrease regularly on a wide range of services, including computing and database solutions. Joey Parsons, Head of Operations at Flipboard, said his company keeps a close watch on these developments to uncover new opportunities to lower operating costs. His business uses CloudFront, EC2, and other AWS services to scale up and control costs as the Flipboard reader base grows.
Parsons has found that Amazon's frequent price changes can significantly impact the savings available—in a positive way. However, his company does have to constantly reevaluate previous choices and assumptions to take advantage of these savings. For example, with lower pricing, it can make sense to move from a medium to a large solution that is now within the company's budget. A similar reevaluation process takes place whenever AWS brings a new service option to the table. But this assessment requires more than number-crunching. Flipboard has to try things out before they fully commit. "Whenever AWS offers a new instance type, we'll take a service role of a similar instance type and deploy it and see how that instance performs with a portion of our production traffic. If it holds up, it's fast, and the pricing makes sense, we'll switch over an entire fleet of our services to reap the benefits immediately."
With IT moving into its intended role of adding value to the enterprise, it's important not to lose sight of the responsibility to capture cost savings as well. If AWS continues slashing prices, just keeping track of all the ways to save money may turn into a full time job.
How do you optimize your AWS return on investment? Let us know.