One claim SSD manufacturers love to make is that SSDs are more reliable due to no moving parts, however, as Jeff Atwood reported at Coding Horror, they are not as reliable as the claim to be, at least the SSDs he got in 2009.
Of three Crucial 128GB SSDs he bought in October 2009 for his Stack Overflow team, two have since failed and his friend Joel also had an Intel SSD fail in its ThinkPad laptop bought about the same time. At first this just seemed like a coincidence until he found out that a company friend Portman Wills bought 8 SSDs over the past two years and all 8 have since failed! These included SSDs by Super Talent, OCZ, G.Skill, Crucial and Intel.
As the blog mentions when an SSD fails, it is catastrophic in that one moment it is working fine and then in an instant, it has failed where no data can be retrieved from it. This is what I experienced when I had an OCZ Agility 60GB fail. I left the room with the PC running idle and came back with a “Disk boot failure” on the screen, as the SSD disappeared from the BIOS, causing the OS to panic, reboot and then get stuck at the BIOS boot error screen since the SSD was the only bootable drive.
While it is not pretty to experience a drive failure, this certainly did not put the blog writer or its friends from getting new SSDs, as once one has used an SSD, it is hard to go back, much like going back to dial-up after a DSL modem fails or back to using payphones when the mobile goes on the blink. In fact, Jeff says he has since bought an OCZ Vertex 3 and is willing to stick with SSDs even if it means having an SSD fail every 12 months on average!
In my desktop configuration, I have my OS and applications installed on the SSD and my user profile redirected to a hard disk, where I also store all my photos, data, etc. Every week or two, I attach an external hard disk and mirror the hard disk over. I also have Windows converted to make an OS partition image backup to the hard disk every Sunday. So when the Agility SSD failed, I simply ran the Windows 7 restoration process on a new SSD and continued where I left off as if nothing happened. As my user profile was redirected, I did not lose any documents, photos, etc. from the SSD failure. Obviously with a laptop, one needs to be careful to make regular backups, where the SSD is the only drive in the laptop.
My gut feel, having tracked this for a while, is that early 2010 was the reliability inflection point for most SSDs. It was then that most SSD controller chip makers figured out how to create controllers that didn’t degrade (remember the 3 second pauses that early SSDs used to have?), and would work more reliably. I’d be interested to hear about SSD failures for disks purchases in 2010…