Does AWS really save money?
AWS is not just one thing. It is many, many services. Whether you save money or spend more depends on what and how you use these services.
- Are you moving Virtual Machines (VMs) and NVMe disk storage from a well-run on-premises environment to AWS EC2? You will spend more. A lot more.
- Are you architecting an environment to benefit from cloud-native services that scale with demand and turn off entirely when not needed? You can save a lot.
In both instances, savings are rarely the goal – instead, our customers are looking for benefits like increased developer efficiency, shorter time to market, etc.
In other words, you might not save money, but you may make more – if your business can use these cloud benefits to drive more business.
What is AWS savings plan?
There are two different kinds of savings plans:
- Compute, which covers EC2 instances, AWS Fargate, and AWS lambda usage regardless of instance family, size, AWS region, operating system, etc. There are 88,480 different EC2 instance types at the time of writing, 44 AWS Fargate resources, and 80 AWS Lambda covered by a compute savings plan.
- The other kind is the EC2 Instance savings plan, which gives a slightly higher discount for specific families of instances but has far less flexibility.
We recommend the Compute savings plan. You maintain cost control and reasonable flexibility.
How do AWS savings plans work?
In short, a compute savings plan is a commitment to spend a certain dollar amount per hour on compute resources – both AWS EC2 instances, AWS Lambda, and AWS Fargate.
If you don’t use up the entire amount that hour, it is lost. If you use more than you have committed, you pay the Pay-As-You-Go rates.
When it comes to prioritizing which services should benefit, AWS automatically applies the committed amount to the resources with the highest discount percentage first – within the account with the savings plan.
Once it has covered every computing resource in the account, it spreads it to other accounts, again with the highest discount first.
Therefore, a savings plan should live in a separate account with nothing else in it – that way, it will always cover the tenant’s (or management account’s) use with the highest discount rates first.
Why is AWS so expensive?
A combination of several explanations:
- Firstly, it is value-based pricing, not commodity pricing. If you are coming from on-premises hardware where vendors compete with virtually identical and interchangeable servers, you’re used to cost-plus pricing. With value-based pricing, vendors price according to the value you perceive to get from the solution. This price will almost always be significantly higher.
- Secondly, your switching costs are high, so you are less likely to leave, even as prices stay high. If you have built your solution using AWS exclusive services, where can you feasibly go? Even Kubernetes workloads are implemented slightly differently from cloud vendor to cloud vendor.
- Thirdly, because competition hasn’t really materialized yet. Yes, there are other cloud vendors like Google Cloud Platform, Microsoft Azure, and Oracle Cloud Infrastructure, but, right now they are all just fighting to keep up with demand from customers – not yet fighting for customers. Many more companies and development teams are looking to put workloads in the cloud than people who can help them do it. We do not expect to see fierce price competition any time soon – and it will continue to be limited by the high switching costs.
All that to say, AWS is priced the way it is because they can, and they are a for-profit company.
Any other Questions?
If you have other questions about AWS and how to control your costs, please ask them below or continue reading our blog post on “How to take control of your Cloud Costs“.