Book recommendations for ML/DL Infrastructure Design

I'm looking for recommendations of books in the area of Infrastructure/SRE/DevOps but oriented to DL/ML. I know a few of them that seem to be too focused on production but what about training? How to design a cluster for deep models training? What about caching data for very large datasets? Is there such a thing?

Or maybe a more general oriented book which principles can be applied easily to designing DL systems for training.

Thanks in advance!