Hey everyone! I'm diving into AWS for the first time and could really use some guidance. I'm working on a project that requires a few different types of servers: 1) I need a server that runs 24/7 for some basic calculations (ideally free). 2) I also need another server for heavy GPU calculations that runs once a day. 3) Finally, I'm looking for a database server that can store about 50GB of data, which includes a queue from server #1, results from server #2, and some metadata. I'm trying to keep costs low since I'm still experimenting. Any tips or advice would be super helpful! Thanks!
2 Answers
For your 24/7 lightweight calculations, you can use the AWS Free Tier EC2 micro instance, but keep in mind it's only free for the first year. For your daily GPU jobs, there aren't free options, but using EC2 Spot Instances with GPUs is a budget-friendly alternative. Just run your job and shut it down to save costs later. As for your storage needs, DynamoDB's free tier covers up to 25GB, but for your 50GB, S3 would be a more economical choice. Overall, you're looking at around $5–$20 a month, which isn't completely free but should work for your experimental phase.
It's a bit tricky without specific details, but for your calculations, consider using event-driven approaches with API Gateway and Lambda to scale easily, especially since they don’t have to run constantly. You could also containerize your work with ECS, or stick it behind an EC2 instance set up for autoscaling. For daily heavy loads, AWS Batch is a solid choice. Regarding storage, if you want a traditional SQL approach, go with Aurora; if you're okay with NoSQL, DynamoDB is faster and cheaper.
Thank you! I'll look into everything, needed a bit of orientation to get on track.
Thanks! Needed an overview of what to use/how much it would cost and seems about right, should be feasible.