"I don't think [disk array manufacturers are] going to be forthright with giving people that data because it would reduce the opportunity for them to add value by 'interpreting' the numbers." Steve Smith, independent consultant, former EMC employee
I continue to wonder how NetApp and EMC can continue to charge so much for their reliability reputations, when so much of their 'value add' is riding on suspect reliability ratings of the drives themselves. In light of the studies listed in a Computerworld article , ("Vendor disk failure rates: Myth or metric? Disk problems contribute to 20% to 55% of storage subsystem failures"), perhaps it is time for customers to insist on performance and reliability guarantees? Currently, marketing and sales folks have created a common perception that using RAID and corporate "proprietary software magic' mitigates most issues and makes data loss on enterprise storage rare. But is that enough of a reassurance?
While the reality is that data loss on these systems is quite rare, the ongoing lawsuits and articles about prior art of the underlying software are eroding the voracity of the storage companies marketing stories. People are starting to look under the hoods of these systems, and when they do they find pretty ordinary computers and off-the-shelf-components, with very fancy bezels.
How much more reliable is a brand new storage system than the one made two years ago? What is the MTBF of a system built today compared with one made two years ago? Can the storage vendors quantify their reliability figures and compare them to the competition and to their superseded models? Often customers come to us and ask Zerowait to help them evaluate their costs of upgrading compared to maintaining their older equipment. Together we find that maintaining their legacy storage equipment makes a lot of sense when the performance gains of the new systems are meager. Over the years experience has shown us that new bezels, marketing schemes, and sales slogans don't improve storage performance.