Experience working with a 600 TB ElasticSearch cluster

I love war stories around tools: all the lessons learned and tweaks needed. At scale, it's never, ever as simple as "just use X".