Came across this earlier today and thought I'd share: "A stored SSD, without power, can start to lose data in as little as a single week on the shelf." https://blog.korelogic.com/blog/2015/03/24#ssds-evidence-storage-issues Thoughts? Comments? Not that I'd expect most of us here to be effected, but my laptop (for example) doesn't get powered on nearly as much as my desktop!
Those appear to be the JEDEC requirements for data retention (i.e. NAND must meet those standards to labelled as JEDEC compatible). Unless you happen to live somewhere where the average air temperature is above 30°C, then you can count on your SSD retaining data for a minimum of 1 year without errors.
There's a big discussion about this at Wilders. The important thing to remember is this is about storing the SSD without power - like on a shelf. And 30°C is pretty warm (86°F). Obviously, an SSD is not for archiving historical data, and you should still have a viable backup plan to HD or optical disks, but it is still safe to use SSDs on your computer to store your documents and programs as regular load leveling and TRIM functions will keep the data refreshed.
I disagree. I think they know a lot about technology but they are more interested in sensationalizing headlines, increasing their readers/viewers/listeners, and interjecting themselves into the news for the publicity than they are in following the facts, discovering the truth, and "reporting" those truths and facts as "news". In other words, they think they are entertainers, and not reporters of facts - that is, until someone calls them out, then they scream about their First Amendment Rights (Freedom of the Press here in the US). And it's this sensationalizing and attention mongering that really irks me most about the media (and shysters too!).
Thanks for the discussion on this and the link to the Anandtech article as well as the wilders discussion. Makes sense really that SSD shouldn't be used for long term storage
The biggest problem i had, with how the whole thing was reported, was that it was claimed that the slides were only a couple of months old - whereas, they're from October 2010. Back in the day (as 2009 is when i started really looking at the tech & 2010 when i bought my first SSDs), loads of people (myself included) were well aware of this as the tech was new & we were still trying to actively look at the tech to establish best practices... ...so either the snazzy tech websites that are also reviewing SSDs either weren't aware of it (in which case their expertise is suspect) or they were & simply didn't bother relating it. [NB i would strongly suspect it's the former, as these are the same sites who will typically throw out nothing but AS-SSD & CDM & whatnot results which are meaningless unless you've got a heavy usage type which happens to predominantly use one of the few i/os that are actually tested with them.] But, (as i commented on Anand), rather than sticking up their hands & saying "mea culpa - we should have told you about this 4.5 years ago"... ...they're pretending that it's something new to obfuscate the fact that they either didn't know or didn't tell anyone. Yeah, so my personal opinion is that any site that's presenting it as new news simply don't have the expertise with SSDs - but then there's none that i'd actively recommend to an uninformed consumer beyond Anand anyway. Well, whilst i know what i'm looking at with performance charts & whatnot (& particularly how most of them are completely pointless for a consumer - & even the odd one that is isn't useful for all consumers), not everyone does... ...so for most people, beyond knowing that they don't need the fastest thing out there for playing games & browsing & whatnot, the light or heavy charts on Anand (as the destroyer isn't very useful), plus a quick read through to see if there's any glaring issues (particularly with performance consistency & trim), should be sufficient... Well, it's only if you're going to be regularly doing some time sensitive significant task with very large amounts of either highly sequential or highly random i/os that anything else is really relevant in buying a SSD. Otherwise, it's important to remember that they're just the minimum specs for data retention for the usage types... ...so you'd expect several times the retention stated with current consumer drives (esp at reasonable temps)... ...whilst it's not reasonably the case that enterprise drives are shonky in comparison - simply that both the need for 100% data retention is significantly greater & they are more likely to be put under a far heavier load which will have a detrimental effect on retention as the drives get older - so it's predominantly a 'better safe than sorry' approach for businesses.
The article that seems to have started all this mess is this ZDNet article by Zack Whittaker. He is the irresponsible "journalist"... cough, cough, choke, choke, puke, puke... and editor for ZDNet who claimed (my bold underline added), And then he quoted Alvin Cox's JEDEC document "JESD218A which was written in February 2011 and which cited the 2010 JEDEC standards which talk about these "retention rates". And then, because so many bloggers subscribe to ZDNet feeds, irresponsible bloggers pretending to be real journalists and IT experts simply repeated Zack Whittaker's inaccurate claims, exacerbating and obfuscating the whole issue. This, of course, caused panic among readers who expect responsible and truthful reporting from these people, but aren't getting.
Whilst it's not hugely important, simply as factual corrections then the thing that originally kicked this off was a blog post on KoreLogic which was solely discussing it in terms of evidence for court cases... ...but they didn't state that it was a recent presentation - which was instead an invention/lack of fact checking by Zack Whittaker... ...which then most outlets treated as gospel. Otherwise, Alvin Cox's JEDEC document is dated 4th October 2010 (created on the 3rd, modified on the 4th) - which makes it reasonable to assume that it's the PP slides that were used in the presentation by JEDEC in San Jose on the 5th, which included JESD218.
Looking at time stamps, you are right. But his blog did not seem to stir up the controversy as the ZDNet article did, perhaps because he did not claim it was new research.
i was simply meaning that it was the blog post that kicked off the thing being (mis)reported recently... ...as nothing else at all seemed to have appeared about it within a timescale that's reasonable to have caused the ZDNet thing. Yeah, so it was only a factual statement about what the root cause was for it suddenly appearing as an issue very recently - when the info's been there for >4.5 years. Anyway, it's not that i'm actively disagreeing with you about anything, as you're correct that the blog was reasonably proportionate within the context of securing evidence for legal proceedings (& i didn't say otherwise) - simply that, when i heard about it a couple of days back, i actively looked to see what the timeline had actually been for anyone to pick up on it again after so long.
It may be well-known that flash has limited retention, but how long would you expect? I have the specs for some old flash (in a PIC microcontroller), which states a minimum retention of 40 years from -65°C to +150°C. I know modern flash is going to last less than this, and will only get worse as density increases, but 1 year at 30°C seems... poor.
I do not believe that is a fair conclusion/assumption. All higher densities suggest is more data. And with more data, there is a greater risk of something getting lost. If you write a 1 page document and 10 page document, there is a greater chance you will have more misspelled words on the 10 page document just because there are more words. Plus with SSDs, the makers are fully aware that retention is an important factor in any storage media. So as they advance the density technologies, they are addressing retention concerns too. It is, but again, that was from several years ago - not with current generation products. And again, while certainly some folks may store these devices in 30°C (86°F) environments, most users would not tolerate their homes getting that hot for more than few days out of the year without air conditioning. Not to mention, we really need more information on what that really means. Is that 30°C 24/7/365? What about at night, or during the cooler months?
From the article and comments at Anandtech (emphasis mine): "Remember that the figures presented here are for a drive that has already passed its endurance rating, so for new drives the data retention is considerably higher, typically over ten years for MLC NAND based SSDs." I.e. it depends on how much you abused the drive while it was in use. Then again, if you haven't touched it in years, the data in it is probably not very valuable..
