Well, this hunch has turned out to be correct.
The cloud storage firm has published a reliability study exclusively dedicated to the SSD units of its servers, which show a progression of failures very similar to that of hard drives up to the fourth year. This is when traditional hard drives begin to fail much more markedly (3.55% in the fifth year), while SSDs, devoid of mechanical elements and with 100% digital operation, remain stable (0.92 %).
To standardize the data as much as possible, Backblaze has exclusively used bootable drives.
Of course, Backblaze’s reliability analyzes are derived from observing data in a professional setting and it is unwise to simply extrapolate them to a home setting. All in all, their data is revealing and shows that SSD drives currently make all the sense in the world for servers. At least from a reliability point of view, since hard drives still have the advantages of storage capacity and cost per MB.
Error rates on the different SSD drives used by Backblaze.
The data extracted by Backblaze has been obtained after about five years of use and at first it seems very constant, almost enough to make predictions. However, its evolution in the longer term will have to be seen, since the NAND memory degrades after a certain number of write cycles, and it could be the case that after a while the failure rate of the SSD units shoot. This and other aspects, such as the ability to go beyond the maximum number of writes accepted by the manufacturer of each drive, will be observed in future studies of Backblaze.