Unpowered for a few days, some SSDs can begin to lose data

[watercooled]

I've just pulled up the source PDF and it doesn't seem entirely clear what those tables refer to. Taking the surrounding slides into consideration, are they actually showing a minimum required specification rather than an actual tested drive?

A lot of the headlines for this story seem overly dramatic, and possibly taking the pdf out of context. I wonder if anyone's actually asked JEDEC to clarify, as it might be the case that these are slides meant to accompany a spoken presentation so lack details.

[watercooled]

I sent a quick email directly to JEDEC asking for some clarification on this issue and received a reply linking me to this article: http://www.pcworld.com/article/29251...after-all.html

[leexgx]

one thing this article and other places that are posting this information about short life of NAND is this is when the NAND is at the end of the spec life (more applicable to TLC then MLC due to the 1000-3000 TLC vs 10,000 MLC write cycle life) but your hammering the SSD to past its spec life of 1000-10000 then you can have issues

say after 300-1500TB has been Written is when you can have problems if data loss when the drive is off for extended time and is in a warm environment

ignore the 30-100TB written data endurance as that is for warranty so you don't use it in server workloads as most ssds can do 500-1500TB of written data before they have issues with NAND no longer retaining data and the firmware having to take action to retire pages or they fail outright (the drive would be flagging warning as its life will be below 10% even windows will be telling to to replace the drive)