Information technology has certain features that make it possible to calculate probable dates of demise. It’s all digital, with a finite number of bits and bytes, and calculable error rates. As disk storage capacities increase, technologies viable today may run out of steam tomorrow. They cannot scale forever. Unlike vinyl records in the music industry or Polaroid cameras (a bit of cheat) that were written off, but then experienced resurgence in their markets, when a disk drive is dead, it’s dead. Here is the thinking behind the disturbingly precise estimate that by 2019, RAID 6 drives should no longer be part of the IT landscape or the disaster recovery scene.
The basic principle of RAID 6 drives, similar to RAID 5 drives before them, is to store data with extra parity bits added in. If one of the disks in the RAID box stops working, the data on that disk can be reconstructed by using the data off the remaining disks, together with the parity bits. In a common RAID 5 configuration involving 7 disks each with 2 terabytes of storage, it can be calculate that one failed disk will still lead to a 62% chance overall of data loss. RAID 6 handles the problem by storing more parity data. There is a performance overhead, but the upside is increased levels of data protection.
The flies in the ointment as drive capacity increases are however numerous. They include longer time to rebuild data, more errors hiding on bigger disks and greater chances of a second disk failing in the RAID 6 system that now uses 14 disks where the comparable RAID 5 system used 7. Further calculations have been made and curves plotted. They show RAID 6 probability of data loss increasing with total storage capacity to reach a level in 2019 that is the same as the one at which vendors previously stopped recommending RAID 5 drives. Now, if only other future disasters like snowstorms, bush fires, floods, lockouts or even virus attacks could be predicted with the same confidence…