Use cases for clusters with slow connection

  • Medium: It contributes to significant difficulty to complete my task, but I can work around it.


Let’s say I have a network of pretty strong computers, but they can talk to each other only via the public internet, we can assume they have “pretty OK internet provider”.

I’m looking for ray-based projects that would make sense in this setting. IOW: things that could use a lot of computing power, but don’t require sending giant amounts of data between nodes in a cluster.

I was thinking of:

  • Hyperparameter tuning
  • Data preprocessing
  • Data generation

But these are general topics, and I’m looking for something more specific. Perfect answer would be “hey look at this cool project, it will work just as well whether or not you have a fast connection”.