Overall, using Redis, Aurora Postgres, Lambdas written in Golang, and AWS EventBridge allowed us to build a stable and robust system that could easily scale to meet our needs. I highly recommend these technologies to any startup if you have a highly proficient team of developers. And if you are looking to build a reliable, scalable system from the first MVP release, faster and with scale in mind.
Redis was a great choice for our caching layer due to its speed and scalability. It allowed us to cache frequently-accessed data, reducing the load on our database and improving performance.
We chose Aurora Postgres for our database because of its high availability, durability, and scalability. This allowed us to easily scale our database as our data grew, while ensuring that it remained reliable.
Lambdas written in Golang allowed us to easily build serverless, event-driven microservices that could scale up or down based on demand. Golang is a great language for building scalable and lightweight microservices, and Lambda is a perfect fit for running these services in a serverless environment.
Finally, we used AWS EventBridge to manage the events between our microservices, decoupling our system into smaller, more manageable pieces. EventBridge allowed us to easily trigger actions based on events, making our system more responsive and improving performance.
For my next project I'll definitely use functions / serverless lambdas but will give an opportunity to Temporal.io as an alternative to AWS Step-Functions looking for a more robust and faster to grow code base. Temporal.io is a Single stateful service in your entire deck stack that lets everything else be stateless... Which as been one of the main pain points using AWS lambdas a part from the cold starts and few other issues we had to face on this two years learning journey.