Putting ‘AWS Savings Plans’ to work
In continuation of my previous blog (The Art of Choosing), where I discussed the evolution of various AWS pricing models, this week, the focus is on comparing the various AWS Savings Plans available and putting them to work.
The baseline for exploring the pricing models is to understand the relationship between the nature of workloads and the choice of AWS instances. Studies reveal that:
- For Predictable Mission-critical workloads, customers choose the cost-effective and flexible pricing solution with ‘Reserved Instance.’
- For Unpredictable Mission-critical workloads, the common choice is ‘On-demand Instances’ because of the freedom to turn them on and off as per the workload and pay as per the utilization.
- For Cyclical workloads, customers would prefer ‘Spot instances’. Of course, for urgent computing needs of additional capacity, spot instances are used in conjunction with On-demand/Reserved instances.
But, few workloads do not qualify under any of these buckets, and for lack of a better term, let us identify them as “Semi-predictable workloads.” The Semi-predictable workloads have a consistent compute usage to a threshold point beyond which they vary. We must recognize these workloads and map the appropriate Savings Plans.
Amazon has announced the new flexible pricing model “Savings Plans” on 6th Nov 2019 which allows customers to save up to 72% on Amazon EC2 and AWS Fargate in exchange for making a commitment to a consistent amount of compute usage (e.g. $10/hour) for a 1- or 3-year term.
Every type of compute usage has a Savings Plans rate and On Demand rate. You will be charged the Savings Plans rates on the committed usage and any usage beyond the commitment will be charged at regular On Demand rates.
There are two types of Savings Plans:
- EC2 Instance Savings Plans (ISP)- This plan is similar to Standard RIs which offers higher savings, up to 72% discount, but requires the users to commit to the usage of individual instance families in a region. For example, m5 in US East (Ohio). The discount will apply to any instance usage within the m5 family within that region regardless of size, OS, and tenancy. Conceptually, it is similar to an instance size-flexible Reserved instance regardless of OS licensing concerns.
- Compute Savings Plans (CSP) – This plan is conceptually similar to Convertible RIs which offers discounts up to 66% with higher flexibility (no need to commit to an individual instance family within a region), and discounts will apply against any EC2 and Fargate compute usage regardless of instance size, family, OS, tenancy, or region.
Each of the AWS Pricing models has relative strengths and capabilities towards cost optimization. The below table helps you choose the best option, which further enables you to get maximum savings with reduced risk.
Below you can see the Savings Plans rates and On Demand rates for every type of offering. Few sample screenshots:
Compute Savings Plans for Amazon EC2
Compute Savings Plans for AWS Fargate
Compute Savings Plans for AWS Lambda
EC2 Instance Savings Plans
Here are few guidelines while choosing the pricing model , especially related to Savings Plans
Over the last decade or so, the cost optimization world is advancing in multiple ways to provide flexibility and discounts to save costs. The new AWS ‘Savings Plans’ is a massive enhancement over Reserved Instances. While organizations still purchase Reserved Instances on case to case basis, the relative simplicity and flexibility of Savings Plans will make a difference to attract many organizations.
While there are many pricing options available under the AWS umbrella, customers still need to evaluate how they consume it. Other aspects of optimization, such as Monitoring, Rightsizing, Auto-scaling, Spot instances, Lightsail, S3 and Containers play a huge role in optimizing costs.
Achieving cost efficiency is an art, so customers should be wise enough to use advanced knowledge of AWS cloud-native tools in coordination with experts who can take advantage of all the relevant services provided by AWS.