ROC Part 2: ACME Benchmarks (so far)
Race to recover vs. race to finish line
Many opportunities to compare commercial products and claims, measure value of research ideas, … with availability benchmarks
Maintainability benchmarks involve people, but so do most research by social scientists
Partial failures: Evaluate “Service level” benchmarks that insert faults that do not bring down entire service for all users?
Even initial Availability benchmarks find peculiarities of systems measured
Lots of low hanging fruit (~ early RAID days)