Tom’s SSD study headline

Author Andrew Ku digs into his subject in great detail and with considerable gusto, so I strongly urge interested readers to dive in and work through all 9 pages of the original story. The focus, appropriately enough for readers of this blog, was on Intel SSD failure rates in scenarios that spanned use in the datacenter all the way to desktop and laptop PCs. Ku produces some very interesting findings (see page 9 of his story for the full list) that include the following elements:

• Annualized failure rates exceed manufacturer’s claims.
• Drives are less like to fail in the first year of use than often reported; in fact, failure rates increase steadily with age.
• Failure rates for consumer and enterprise drives are nearly identical.
• Data redundancy with SSDs doesn’t have to be expensive: use continuous backup to conventional disks to ensure availability, and forget RAID.
• SSDs fail at only slightly lower rates than conventional HDs

The bottom line is that rumors that SSDs are less reliable than HDs are not born out by the study data, but if SSDs are more reliable than HDs, the difference is too slight to justify their considerable cost differential in and of itself. Rather, speed, power consumption, and compact form factor seem more likely to count for at least some applications where SSDs are taking over from HDs, particularly for system/boot drives, and in notebook PCs.