Amazon Simplifies Discounts on Reserved Instances

Amazon Web Services has simplified the way it offers cloud discounts to EC2 users that reserve cloud compute capacity in advance.

Instead of providing different level of discounts depending on how heavily the reserved instances are used, AWS now offers a single type of reserved instance discounts. With the new model, size of the discount varies depending on whether the user pays for all reserved capacity upfront, partial capacity, or does not pay upfront at all.

Users can reserve cloud VMs for one year or three years. Savings range from about 30 percent to 75 percent, depending on instance specs and length of the reservation.

AWS has been offering discounted reserved instances since 2009. Its major rivals in the public cloud market, Google and Microsoft, have each done different things with commitment discounts.

Google doesn’t offer cloud discounts on reserved instances on its Compute Engine. Instead, it offers a “sustained use” discount, which is applied automatically once a user runs an instance for longer than 25 percent of a billing cycle. If you use an instance throughout the entire billing cycle, your net discount on it will be 30 percent.

Google introduced sustained use discounts earlier this year. It was a much simpler discount system than Amazon's reserved instance model was.

Microsoft Azure announced discounted commitment plans in 2013 but stopped offering them earlier this year. It still provides the discount (20 percent to about 30 percent) to users that subscribed before it nixed the plan.

Big cloud providers have been battling it out by continuously slashing the rates they charge their users. Rewarding users who make upfront commitments to their services plays a role in that battle, but also has another purpose.

Cloud discounts on reserved instances are a way to court long-term users to the public cloud service whose biggest appeal is the ability to use it temporarily and pay only for what you use. It is an alternative to making heavy investment in hardware and data centers.

Here’s a break-down of the new model from Tuesday’s blog post by AWS Chief Evangelist Jeff Barr: