Editor of the Sanity engineering blog here. Sharing this post from our SRE team about how we reduced the hosting cost for Grafana Mimir while keeping redundancy for zonal failures.
Happy to answer questions, and I'll pull in the author for anything I can't.