Carlos E. R. wrote:
On large installations such as those you and Kai mention, are SMART tests useful to predict failure?
When SMART tells that a disk will fail, that's a very good prediction. We don't collect data how many disks fail without SMART alerts before. But the recent Google paper on disks has data on that. In my client's environment (one of the world's largest automotive R&D sites), all server disks are RAID-1. Exchange disks are on-site and are replaced daily. SMART alerts are actually not so of interest for us, disk mirroring and good operating processes are more important. FWIW, important data is on EMC storage boxes. For SAN products, they have the highest reliability that we found. And the ability to do (asynchronous) SRDF mirroring across sites is good for some disaster recovery use cases as well. (Though not practical for the _real_ important data, there one needs application-level delayed data replication; but that's another topic.) Joachim -- =-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=- Joachim Schrod Email: jschrod@acm.org Roedermark, Germany -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org For additional commands, e-mail: opensuse+help@opensuse.org