Hey everyone! I hope you're all doing well. I've got some experience with Kubernetes, having used EKS and AKS for fun projects, but now I'm stepping up my game for a production deployment. My current app is quite demanding—it maxes out my PC's resources when I run it in Docker. I'm looking for advice on what type of VPS and its specs I should consider. Here's what I'm working with:
- 8 Spring services
- 2 MongoDB instances
- 1 RabbitMQ instance
- 3 PostgreSQL instances
- 1 Ollama instance running Mixtral 1.5
- 1 Chroma instance
I know it's tough to be precise, but I'd appreciate any ballpark estimates. Thanks a lot!
3 Answers
You should have some baseline from your local runs. If it’s already running and you know how much it’s using, that can help. Just remember, the real test will be what your app needs when it scales up. A more powerful VPS will likely make a big difference for resource-heavy processes like Mixtral.
One approach could be to check how much resources your app is using right now. That can give you a rough idea of what to aim for. It might be a good starting point even if it's currently maxing out your local setup.
Yeah, I considered that too, but my app really pushes the limits, maxing out 16GB of RAM and all CPU with terrible response times. So, it’s not the best metric to go by.
Keep an eye on which part of your app is consuming the most resources. It seems like Mixtral could be your main consumer, and the other components might just be sharing the leftover resources.
Definitely! I’ve had similar setups, and Mixtral was a significant resource hog for me too. Make sure to leave enough capacity for your other services!