Confused about AWS EC2 instance types and AWS EC2 pricing options? In this post we explain some of the ways that it is possible to reduce cloud costs, with a useful video at the end that provides figures demonstrating the cost range for a specific instance type.
Instance Types and AWS EC2 Pricing Options
AWS offers many different instance types, each suited to a particular range of uses. Within each of those instance types there are several different pricing options, ranging from the most expensive ‘On Demand’ options to Reserved Instances, Convertible Reserved Instances, and Spot Instances, all of which are available at a discounted price.
On Demand instances are the most flexible and naturally the costliest option. As the name suggests, they are purchased on demand when they are needed. They can be turned on and off whenever required, spun up and run for an hour or so before being terminated, or they can be run round the clock. They will always be there when you need them. If you switch them off, you will be charged for the disk, but you will be able to save a considerable amount as you will not be charged for the CPU or memory.
Users who are able to commit to usage over the long term can achieve considerable cost savings by purchasing Reserved Instances. Users simply choose their instance size and then commit to that instance for a year or three years, with options available to pay monthly, every six months in advance, or for the entire duration of the contract in advance.
The cost savings can be considerable compared to On Demand instances, although to achieve the cost benefits the instances need to be utilized throughout the term of the contract. Typically, the cost benefits of a 1-year contract are not realized until at least month 10.
Convertible Reserved Instances allows users to have the cost saving benefits of Reserved Instances, although they incorporate greater flexibility as they can be exchanged for other (more expensive) instance types if required. They are cheaper than On Demand, but not as cheaper as the inflexible Reserved Instances.
Another alternative is the use of Spot Instances. These are On Demand Instances that are sold off at a significant discount. The cost of Spot Instances is variable, as is availability, and there are no guarantees of availability. There can be interruptions and the instances can be switched off without notice. There is typically a 15% chance of interruptions, so they are not suitable for all applications and processing, although the cost savings are significant. Spot Instances are typically available at around 65% of the cost of On Demand instances.
In this useful video, Chris Parlette, the Director of Cloud Solutions at ParkMyCloud, explains the AWS instance types, the benefits of each instance family, and the different costs of using each and areas where savings can be made.