If this is your first visit, be sure to check out the FAQ by clicking the link above. You may have to register before you can post: click the register link above to proceed. To start viewing messages, select the forum that you want to visit from the selection below. |
|
|
Thread Tools | Display Modes |
#31
|
|||
|
|||
ssd defrag
In message , Paul
writes: Bill in Co wrote: Has it gotten to the point now that SSDs are considered to be just as reliable, long term, as the standard hard drives, even with all the consequent writes and rewrites (also potentially limiting the SSDs "longevity")? (I mean when used as your main drive)? But maybe SSDs still haven't been out quite long enough to yet assess their long term reliability and longevity. An excellent question that is not often enough asked, and certainly even more rarely answered! It's gotten to the point you can use them. Hmmm ... They don't insta-brick like they once did. What you say below tends to counter that - or, perhaps, you meant they don't _unpredictably_ do so. The user "John Doe" had one insta-brick on him. They're still potentially susceptible to power events. Check the SMART table, to see if "the drive thinks you've been abusing it". There's a field for that (abrupt power loss). For example, even if I safely remove an SSD connected to a USB to SATA 2.5" adapter, the SSD counts my unplugging the cable after Safely Remove as an abrupt power loss. It should not do that, if the command was making it through the protocol layers properly. (The drive should have been placed in a "spun down" state.) Not good (-:! [I take it this "abrupt power loss" counter "punishes" "abuse".] You still need to back them up. As with anything. Don't leave your data files on one. Leave your OS on the SSD, move your data files to the HDD. The "end of life" of an HDD today, is much more gentle than the "brick state" an Intel SSD drive enters at the end of its wear life counter. That's what's kept me on HDD (that plus the fact that most laptops only have one drive bay anyway). [I do _partition_ it into C: (OS and software) and D: (data).] Intel will allow neither read nor write, when the computed amount of write cycles is exceeded. Has anyone asked them (and got an answer) _why_ they do that? Blocking writes OK, but why block reads? Samsung will likely allow the drive to continue, so you could, say, do a last backup. Intel SSDs don't allow even that. Dig a hole in the back yard, and throw your Intel SSD in the hole, when that happens. "No data recovery for you." Always research the "end-of-life" behavior of any SSD you buy, so your backup strategy has you covered. Paul JPG Ever been frustrated that you can't *disagree* with a petition? If so, visit 255soft.uk - and please pass it on, too. -- J. P. Gilliver. UMRA: 1960/1985 MB++G()AL-IS-Ch++(p)Ar@T+H+Sh0!:`)DNAf Grief generates a huge energy in you and it's better for everybody if you harness it to do something. - Judi Dench, RT 2015/2/28-3/6 |
Ads |
#32
|
|||
|
|||
ssd defrag
Paul wrote:
Bill in Co wrote: Has it gotten to the point now that SSDs are considered to be just as reliable, long term, as the standard hard drives, even with all the consequent writes and rewrites (also potentially limiting the SSDs "longevity")? (I mean when used as your main drive)? But maybe SSDs still haven't been out quite long enough to yet assess their long term reliability and longevity. It's gotten to the point you can use them. They don't insta-brick like they once did. The user "John Doe" had one insta-brick on him. They're still potentially susceptible to power events. Check the SMART table, to see if "the drive thinks you've been abusing it". There's a field for that (abrupt power loss). For example, even if I safely remove an SSD connected to a USB to SATA 2.5" adapter, the SSD counts my unplugging the cable after Safely Remove as an abrupt power loss. It should not do that, if the command was making it through the protocol layers properly. (The drive should have been placed in a "spun down" state.) You still need to back them up. Don't leave your data files on one. Leave your OS on the SSD, move your data files to the HDD. The "end of life" of an HDD today, is much more gentle than the "brick state" an Intel SSD drive enters at the end of its wear life counter. Intel will allow neither read nor write, when the computed amount of write cycles is exceeded. Samsung will likely allow the drive to continue, so you could, say, do a last backup. Intel SSDs don't allow even that. Dig a hole in the back yard, and throw your Intel SSD in the hole, when that happens. "No data recovery for you." Always research the "end-of-life" behavior of any SSD you buy, so your backup strategy has you covered. Paul Sounds like the most conservative approach still is to use a regular HDD. (And when you said just leave the OS on the SSD, I'm not sure if you were including the Program Files too, but I'm guessing you were). But the way SSDs fail (as a brick) sure is concerning, at least to me. And I've had that happen on a flash drive. |
#33
|
|||
|
|||
ssd defrag
On 12/4/2018 2:31 PM, Bill in Co wrote:
But the way SSDs fail (as a brick) sure is concerning, at least to me. How old is this anecdotal information? Is it representative of current drive designs? Also, under real world, typical use, how long before a drive exceeds its write cycle limit? Given that SSDs are appearing in more and more systems, including some where it cannot be user replaced, are these fears real or imagined? And I've had that happen on a flash drive. How long ago? Today's products may be different than those from a few years ago. Also, SSDs and flash drives are not designed the same. |
#34
|
|||
|
|||
ssd defrag
joe wrote:
On 12/4/2018 2:31 PM, Bill in Co wrote: But the way SSDs fail (as a brick) sure is concerning, at least to me. How old is this anecdotal information? Is it representative of current drive designs? Also, under real world, typical use, how long before a drive exceeds its write cycle limit? Given that SSDs are appearing in more and more systems, including some where it cannot be user replaced, are these fears real or imagined? And I've had that happen on a flash drive. How long ago? Today's products may be different than those from a few years ago. Also, SSDs and flash drives are not designed the same. It's not that hard to find stories. https://forums.windowssecrets.com/sh...ling-questions 2011 Oct.12 SSDs fail by design, at the end of wear life. If the flash chips have 3000 write cycles (TLC), then the manufacturer does the math, and with wear leveling, most all cells will have been written 3000 times, when the SSD "closes up shop and stops responding". Intel drives brick on both read and write, so you "cannot make a last backup or clone them" at end of life. Some other brands may stop writing, but allow reading. Some brands continue to allow you to do whatever you want, and a famous experiment on the Internet showed that extremely long lives are possible if that feature is not present at all. SSDs fail by accident, due to firmware issues. Some SSDs have triple core CPUs inside, with various maintenance and operation threads running. There can be race conditions in the code. All the code has to be written in such a way as to be "power fail safe". In some cases, the next time the drive starts up, it would have to consult a table of "what we were doing at abrupt power fail" and clean up. A few models of the larger HDDs (maybe 8TB and up, with 256MB DRAM cache) now resort to using Flash when the power fails. The cache is written to Flash, the Flash is transferred to disk on the next power-up. You don't typically have this feature on consumer hard drives. That's to show an example of "delayed cleanup" where extra electronic components were added to make it possible. ******* USB flash drives are different. They were good in SLC and MLC days. Not so good in todays TLC generation. I've lost two TLC drives (32GB) with only light usage over a period of one year. My conclusion is, that they don't use the same wear leveling method as SSDs. Wear leveling is patented, and a license likely costs money. ******* The eMMC drives in tablets, are 32GB in size. Guess what kind of flash they're using. eMMC drives are not the same as SSDs, and generally have a much lower IOP rate. NVMe drives have slightly better IOP than a SATA SSD, but have much higher sustained bandwidth. But they can be made with TLC too. And have decent wear leveling (not like the USB flash drive). The wear leveling isn't necessarily documented in a way that a consumer can benefit from the knowledge. ******* Todays products may be different, but they're not "bullet proof". There are some enterprise SSDs with a single Supercap inside for emergency power. Those handle abrupt power loss a little better than all the consumer SSDs we buy, where the Supercap copper pads on the PCB, have no component soldered in. Of course those consumer drives do the right thing, most of the time. But how many million times can you drop the power, without an issue. ******* The good news, is firmware based brickage (something seen in the past on *both* HDD and SSDs), is on the wane. But the other aspects of device design haven't changed. I still recommend an accelerated backup schedule for SSDs, if "you keep your bank statement on it". If my SSDs here tip over, there nothing of value on them (Win10) :-) Paul |
Thread Tools | |
Display Modes | |
|
|