In short, it’s really fast. AWS PostgreSQL for $88 per month can create around 370 records per second in a heavy-indexed table suitable for storing banking transactions. In a simple, almost no-indexed table, the speed is 490 rps.
In this article I will share the details of my experiment. I will show you what test stand I used and the exact table configuration I created for the tests. Also, you will know how quickly DB queries were executed on 15M+ data sets. Hope, you are already excited!
Amazon cloud proposals start from x2 vCPU and 8GB memory installations, and that’s the minimum you can get. I focused on the initial setup, the one marked as db.t3.large. It’s $88 per month for 30GB disk storage and 80% CPU usage (US region).
Below is the test stand I used in my experiment. The Loader app is the one that connects to a PostgreSQL instance over the local network and executes “insert” statements directly. I used a simple EC2 instance: t4g.micro — that’s x2 CPU and 1GB memory.
Dummy data was generated using withshepherd/faker library. The loader can generate up to 3.8k rps locally, without connection to a database. The average ping between the application and the database is ~0.151ms.