Large companies began to actively use SSDs in their data centers, which is why there is not much information about how these drives behave for a given usage scenario.
Google is using SSD in their data centers, and at FAST 2016 conference, they presented a report (p. 67-80) on the reliability of various solid-state drives in continuous operation as a part of the data center. The report presents data on the SSD-drives from 10 different models of corporate and consumer classes that use three types of memory (MLC, eMLC and SLC), and a combined total of more than 1 million days during 6 years.
The study yielded some very surprising results. So, it turned out that the reliability of solid-state drives affects their age, rather than the intensity of use. More emerged drives high-end SLC memory is more reliable than disk drives with MLC memory. Failures SSD-drives are less common than hard drives, but SSD higher rate of occurrence of uncorrectable errors. Therefore, the researchers in the case of SSD recommend more frequent files backup.
An important discovery was that described in the documentation SSD coefficient uncorrectable bit error rate (UBER - Uncorrectable Bit Error Rate) has no practical significance and does not affect the operation of the drive in the future. Bit Error Ratio (RBER - Raw Bit Error Rate) does not increase as fast as we can expect, and does not depend on UBER or other failures.
Failed (bad or bad) blocks are characteristic of many of the new SSD-drives. In 30-80% of solid-state drives appear in at least one bad block in the first four years of use.