Live data from Hacker News

Everything I know about SSDs (2019)

kcall.co.uk

31–40 of 53 posts

Re: Everything I know about SSDs (2019)

#31

Something useful to know that wasn't mentioned: SSDs will corrupt data if sitting for extended periods without being powered on and thus should never be used for cold storage. "There are considerations which should be made if you are planning on shutting down an SSD based system for an extended period. The JEDEC spec for Enterprise SSD drives requires that the drives retain data for a minimum of 3 months at 40C. This…

Note that retention increases exponentially with decreasing temperature, so literally "cold storage" might actually be OK for SSDs.

The retention also goes down with erase cycles, and is usually specified after the rated number of cycles have been reached; I expect those same SSDs to hold their data much longer than 3 months if they're still nearly new.

I still have USB drives over a decade old, but their contents are still intact. Then again, those haven't been written to much, and are SLC and older 2-bit MLC.

Re: Everything I know about SSDs (2019)

#32
post #5

I always thought 'flash' was a holdover from early reprogrammable ROM technology, non-volatile memory that you could erase by literally flashing a literal flashbulb over a little window on the chip. I would've sworn in a court of law that I recall this being called "flashable" memory, that erasing it was called "flashing" it, and reprogramming it in general was called "reflashing" in a sort of synecdoche. And I'd hav…

/pedantic

ROM - Read Only Memory - programmed at the time of manufacture.

EPROM - Erasable Programmable Read Only Memory (AKA UVEPROM because they could be erased using UV light over a period of time, and they had a quartz window to admit the UV.)

EEPROM - Electrically Erasable Programmable Read Only Memory. IIRC it required a special device to erase.

It's been a while since I worked with this stuff but I don't ever recall hearing it called flash.

Re: Everything I know about SSDs (2019)

#33

Something useful to know that wasn't mentioned: SSDs will corrupt data if sitting for extended periods without being powered on and thus should never be used for cold storage. "There are considerations which should be made if you are planning on shutting down an SSD based system for an extended period. The JEDEC spec for Enterprise SSD drives requires that the drives retain data for a minimum of 3 months at 40C. This…

I ran into this for the first time a couple weeks ago. I tried to boot a system from an SSD that had sat for about 9 months at normal room temperature w/out power. It stumbled badly, repeatedly. I booted a live USB, ran non-destructive `badblocks` scan on it and reinstalled the OS. It's been working fine since.

I thought SSDs would last longer than 9 months w/out losing data when not powered.

Re: Everything I know about SSDs (2019)

#34
post #26

Earlier quoted context omitted.

Makes me wonder if those external SSD drives made by Samsung and WD/Sandisk take that into account and use different flash memory with better longevity without power.

If anything, I would expect consumer SSDs to use the cheapest/lowest grade of flash available.

Small USB flash drives get the worst NAND, followed by memory cards, then consumer SSDs, then enterprise SSDs. "Portable SSDs" that are physically much larger than USB thumb drives usually contain standard consumer SSDs in mSATA or M.2 form factor, plus a bridge chip.

Re: Everything I know about SSDs (2019)

#36
post #5

I always thought 'flash' was a holdover from early reprogrammable ROM technology, non-volatile memory that you could erase by literally flashing a literal flashbulb over a little window on the chip. I would've sworn in a court of law that I recall this being called "flashable" memory, that erasing it was called "flashing" it, and reprogramming it in general was called "reflashing" in a sort of synecdoche. And I'd hav…

I never heard of using a literal flash to reprogram EPROMs, but this wikipedia entry[1] makes your story for the origin of the term “Flashing” seem likely EPROMs are easily recognizable by the transparent fused quartz (or on later models resin) window on the top of the package, through which the silicon chip is visible, and which permits exposure to ultraviolet light during erasing. [1] https://en.m.wikipedia.org/wik…

Heard from a greybeard that they had a demo night in university (this was probably early 80s), and one of the demos was was some sort of path-finding rover robot. Of course, it ended up being one of those projects that had show stopping bugs up until a few hours before showtime.

During demo night, it was a big hit, until a stray camera flash got (un)lucky and wiped the microcontroller's EEPROM...

Feels a little apocryphal (I'd assume most flashes have / had UV filters?)

Re: Everything I know about SSDs (2019)

#37
post #10

I've heard SSDs are more likely to "fail fast". Can anyone recommended utilities that monitor and warn before SSD failures?

For NVMe, if you get the SMART data with smartmontools/smartctl, you can inspect Percentage Used. "Percentage Used: Contains a vendor specific estimate of the percentage of life used for the Endurance Group based on the actual usage and the manufacturer’s prediction of NVM life. A value of 100 indicates that the estimated endurance of the NVM in the Endurance Group has been consumed, but may not indicate an NVM failu…

>if you get the SMART data with smartmontools/smartctl, you can inspect Percentage Used.

CrystalDiskInfo[1] can be used for this purpose over on Windows. Some vendor-provided utilities like Samsung Magician will also provide this data with appropriate drives.

[1]: https://crystalmark.info/en/software/crystaldiskinfo/

Re: Everything I know about SSDs (2019)

#38

I've heard SSDs are more likely to "fail fast". Can anyone recommended utilities that monitor and warn before SSD failures?

One surprising feature of "enterprise" class drives is that they often have a faster fail than their consumer class counterparts. the idea being an enterprise class drive will be found in a drive array, and the array will be much happier with a hard fast final failure than a drive that tries to limp along. While the consumer with their single drive is a lot happier when their failing drive gives them every chance to get the data off.

Re: Everything I know about SSDs (2019)

#39

Something useful to know that wasn't mentioned: SSDs will corrupt data if sitting for extended periods without being powered on and thus should never be used for cold storage. "There are considerations which should be made if you are planning on shutting down an SSD based system for an extended period. The JEDEC spec for Enterprise SSD drives requires that the drives retain data for a minimum of 3 months at 40C. This…

Note that retention increases exponentially with decreasing temperature, so literally "cold storage" might actually be OK for SSDs. The retention also goes down with erase cycles, and is usually specified after the rated number of cycles have been reached; I expect those same SSDs to hold their data much longer than 3 months if they're still nearly new. I still have USB drives over a decade old, but their contents ar…

Those are drives that weren't plugged in for a decade? How do you know the contents are intact?

Re: Everything I know about SSDs (2019)

#40
post #17

Earlier quoted context omitted.

EPROMs were erased by many minutes of exposure to UV light, not a flashbulb, in a device called a UV eraser. I've never heard anyone refer to any operation on an EPROM as "flashing".

Yep, "UV eraser" confirmed by an elmer I know. A unit with a drawer and a timer knob, he says you'd typically set for 30 minutes. He says he does recall people calling it "flashing," but not until much later, by which time it would have been actual "flash" memory. As for my own memory, I'm going to file this under Mandela Effect, cross-referenced under Things People Probably Told Me That I've Believed Since Before Th…

Note that the Internet isn't always right, and sometimes the dead trees are the only reliable source.

The Mandela Effect is just a name we've come up with for identifying biological memory faults.

Post reply on HN