Hello everyone, I want to compare distributed file systems and at the moment I am looking a the fact of Single Point of Failures. I found several articles about Lustre which mentioned that Lustre is excellent for HPC (High Performance Computing), but that it has its weaknesses concerning redundancy. Is is true? Or let me ask these questions: In a common Lustre setup, how many MDSs and MDTs will you have? One each? What if this single MDS/MDT fails? Thanks in advance! Pascal