I was writing an article for Infoq on the topic less spoken costs of managed databases and one question that the editor asked is how much margin RDS makes compared to running a Postgres instance on EC2? That intrigued me and honestly, I never did the math so far. I have been using AWS-managed databases since 2016 and I thought what’s a better time to do an analysis of Cloud cost than today ?!
I started by comparing instance types between EC2 and RDS. As far as I understand, these are the same VMs (EC2) that are powering RDS. Anyone with the knowledge, correct me if I am wrong.
When I saw more than a 100% margin, I wasn’t sure whether I was making any mathematical errors. My instant reaction was perhaps there is some difference in other systems resources. I looked at network performance which is up to 12.5 Gigabit, it’s 4 vCPUs with 16 GiB memory. Same-to-same. My understanding is that the newer generation of instances are generally cheaper. However, that’s where the margins are also higher.
So far I analyzed only the instance type. Next, I looked into the storage cost (EBS). These are compared by IOPS, throughput, and no snapshots enabled.
Looks like gp3 is an anomaly. I used the AWS cost calculator and it had a few additional notes on that.
Certainly, there is operation value adds that RDS provides - automated backup, and failover in the case of AZ (though it’s not an apple-apple comparison). So there is some value added for the markup. However, features like performance insights, and RDS proxy have to be paid separately.
Open questions:
[1]. What is the difference between a vanilla EC2 instance and an RDS instance?
[2]. Is there any free tier in GP3 volume type for RDS?
[3]. What’s the markup in GCP and Azure?
References:
https://instances.vantage.sh/rds/?cost_duration=monthly
https://instances.vantage.sh/?cost_duration=monthly
https://calculator.aws/#/createCalculator/RDSPostgreSQL
https://rbranson.medium.com/rds-pricing-has-more-than-doubled-ef8c3b7e5218