Personal computing discussed

Moderators: morphine, Steel

 
just brew it!
Gold subscriber
Administrator
Posts: 48246
Joined: Tue Aug 20, 2002 10:51 pm
Location: Somewhere, having a beer

Re: SSD reliability

Mon Apr 17, 2017 5:15 pm

Random bad sectors are a result of isolated damage to the surface of the platter. A bad head will render an entire side of a platter inaccessible (which will look like periodic runs of bad sectors spread out across the entire disk).
If the world isn't making sense to you, you're either drinking too much or not drinking enough.
 
bfg-9000
Gerbil First Class
Posts: 152
Joined: Tue Mar 01, 2016 9:17 pm

Re: SSD reliability

Mon Apr 17, 2017 5:18 pm

Most of those drives in the endurance tests indeed went read-only, but then bricked themselves after a power reset by design (I have no idea why anyone would ever consider this the proper way to handle running out of P/E cycles). So the data recovery service would actually need a way to read the chips directly, which generally only the original manufacturer (or NSA because they have unlimited money) can do because only they'll know how everything is organized. In comparison a head swap should be doable by any recovery service.

A bad sector can be marked bad by the OS or reallocated to spare sectors by the drive firmware so aren't a big deal. Failed head means the drive locks up when you try to access anything under it. I actually have a few HDD in service with a failed head--I had to partition around the entire bad side of the platter but they still work (the beginning is where formatting crashes and the end is calculated from dividing the total by the number of platter sides). A crashed head is the worst--it can scrape enough material off the platter so the drive sounds like it has sand in it when you shake it, or the bouncing dust quickly sandblasts the other platters so everything is unrecoverable.

It's not just failures--have you ever tried to use an undelete utility on a SSD? Garbage collection seems to be so aggressive that in many cases you only have about ten minutes to use it before it gets zeroed out. Of course after 1759TB it deletes itself!

I'd also like to remind everyone that MTBF is not expected lifespan:
Image
 
just brew it!
Gold subscriber
Administrator
Posts: 48246
Joined: Tue Aug 20, 2002 10:51 pm
Location: Somewhere, having a beer

Re: SSD reliability

Mon Apr 17, 2017 5:25 pm

bfg-9000 wrote:
It's not just failures--have you ever tried to use an undelete utility on a SSD? Garbage collection seems to be so aggressive that in many cases you only have about ten minutes to use it before it gets zeroed out.

With TRIM I am surprised you even have 10 minutes. I'd expect it to be less, since the OS is explicitly telling the drive "please forget about the content of these flash blocks so we can re-use them without forcing you to do a read-modify-write".
If the world isn't making sense to you, you're either drinking too much or not drinking enough.
 
Noldor
Gerbil Team Leader
Topic Author
Posts: 200
Joined: Sun Mar 04, 2007 4:48 am

Re: SSD reliability

Mon Apr 17, 2017 5:34 pm

just brew it! wrote:
bfg-9000 wrote:
It's not just failures--have you ever tried to use an undelete utility on a SSD? Garbage collection seems to be so aggressive that in many cases you only have about ten minutes to use it before it gets zeroed out.

With TRIM I am surprised you even have 10 minutes. I'd expect it to be less, since the OS is explicitly telling the drive "please forget about the content of these flash blocks so we can re-use them without forcing you to do a read-modify-write".


So, the typical undelete utilities one can use on a HDD don't work on SSDs?
 
Waco
Gold subscriber
Gerbil Jedi
Posts: 1849
Joined: Tue Jan 20, 2009 4:14 pm
Location: Los Alamos, NM

Re: SSD reliability

Mon Apr 17, 2017 5:53 pm

Noldor wrote:
So, the typical undelete utilities one can use on a HDD don't work on SSDs?

If you have TRIM enabled, no.

If you don't have TRIM enabled, they're basically the same. Both SSDs and HDDs retain the data you wrote until you overwrite the same LBA or secure erase them. Garbage collection on an SSD will *never* touch blocks that haven't been overwritten.
Z170A Gaming Pro Carbon | 6700K @ 4.5 | 16 GB | GTX Titan X | Seasonix Gold 850 | XSPC RX360 | Heatkiller R3 | D5 + RP-452X2 | Cosmos II | 480 + 240 + LSI 9207-8i (128x8) SSDs
 
Vhalidictes
Graphmaster Gerbil
Posts: 1299
Joined: Fri Jan 07, 2005 2:32 pm
Location: Paragon City, RI

Re: SSD reliability

Mon Apr 17, 2017 5:54 pm

Noldor wrote:
just brew it! wrote:
bfg-9000 wrote:
It's not just failures--have you ever tried to use an undelete utility on a SSD? Garbage collection seems to be so aggressive that in many cases you only have about ten minutes to use it before it gets zeroed out.

With TRIM I am surprised you even have 10 minutes. I'd expect it to be less, since the OS is explicitly telling the drive "please forget about the content of these flash blocks so we can re-use them without forcing you to do a read-modify-write".


So, the typical undelete utilities one can use on a HDD don't work on SSDs?


They *can* if you catch the data before the drive TRIMs it. In most real-world scenarios, undeleting on an SSD doesn't work.
 
Waco
Gold subscriber
Gerbil Jedi
Posts: 1849
Joined: Tue Jan 20, 2009 4:14 pm
Location: Los Alamos, NM

Re: SSD reliability

Mon Apr 17, 2017 5:59 pm

If you don't care about write speeds, you can disable TRIM and have slightly more comfort in your ability to recover files. :)
Z170A Gaming Pro Carbon | 6700K @ 4.5 | 16 GB | GTX Titan X | Seasonix Gold 850 | XSPC RX360 | Heatkiller R3 | D5 + RP-452X2 | Cosmos II | 480 + 240 + LSI 9207-8i (128x8) SSDs
 
ChicagoDave
Gerbil
Posts: 45
Joined: Thu Nov 20, 2014 7:15 pm

Re: SSD reliability

Mon Apr 17, 2017 6:14 pm

A few people have mentioned it already, but for laptops you should absolutely use a SSD due to the lack of moving parts. If you drop or knock a HDD and the head comes into contact with the platter, say goodbye to your data. A SSD can be tossed around with no issues (just like your phone). The speed increase is certainly the most compelling feature overall, but portables in particular benefit from the lack of moving parts.

If your friend really needs extra storage, an external USB drive (SSD or HDD) can always be added. It's really not even something to debate anymore, just get the SSD.

In my personal experience, I have seven SSD drives between my gaming/server, HTPC, laptop and a new external USB SSD. The oldest is a Samsung 840 Pro and I've had zero issues with it (or any of them). Others include a two Crucial MX and BX drives, Samsung 960 EVO M.2, another Samsung 840 pro, and whatever is in my new Lexar external SSD from Costco. All of my computers use them as boot drives, have a second in each desktop for game storage. They're honestly the most important computer upgrade since dGPU's appeared in the mid 90's IMO.
 
just brew it!
Gold subscriber
Administrator
Posts: 48246
Joined: Tue Aug 20, 2002 10:51 pm
Location: Somewhere, having a beer

Re: SSD reliability

Mon Apr 17, 2017 6:37 pm

Waco wrote:
Noldor wrote:
So, the typical undelete utilities one can use on a HDD don't work on SSDs?

If you have TRIM enabled, no.

If you don't have TRIM enabled, they're basically the same. Both SSDs and HDDs retain the data you wrote until you overwrite the same LBA or secure erase them. Garbage collection on an SSD will *never* touch blocks that haven't been overwritten.

Yup. By default (in the absence of TRIM), a SSD emulates the semantics of a HDD. The problem with this is, it results in a performance penalty -- flash blocks have to be erased before they can be rewritten (HDDs do not have this penalty). TRIM mitigates this penalty by telling the drive when blocks no longer contain data that the file system cares about; the drive can then erase these in the background, so that they are ready for re-use.

Modern SSDs do a pretty good job of minimizing the penalty even in the absence of TRIM (and for most workloads they will still outperform HDDs by a wide margin), but performance will always be better if TRIM is enabled.

(It's actually a bit more complicated than described above... erasures need to be done in chunks that are larger than the file system block size, so there's some additional bookkeeping and shuffling around of data involved... but that's the basic concept.)
If the world isn't making sense to you, you're either drinking too much or not drinking enough.
 
Waco
Gold subscriber
Gerbil Jedi
Posts: 1849
Joined: Tue Jan 20, 2009 4:14 pm
Location: Los Alamos, NM

Re: SSD reliability

Tue Apr 18, 2017 8:48 pm

Yep. The deficit isn't huge on modern drives. I run my 8-way RAID 0 without TRIM and don't miss it a bit. If you hammer your drives with benchmarks it's a lot more apparent, but I'd bet that in most user workloads, the difference is little to none.
Z170A Gaming Pro Carbon | 6700K @ 4.5 | 16 GB | GTX Titan X | Seasonix Gold 850 | XSPC RX360 | Heatkiller R3 | D5 + RP-452X2 | Cosmos II | 480 + 240 + LSI 9207-8i (128x8) SSDs
 
CScottG
Gerbil Elite
Posts: 837
Joined: Fri Dec 01, 2006 9:53 pm

Re: SSD reliability

Tue Apr 18, 2017 9:14 pm

Waco wrote:
I run my 8-way RAID 0 without TRIM and don't miss it a bit.


Not often you hear about using SSD's without TRIM these days. I setup my father's system without TRIM - because I had to (given that he's only using SSD's now, and in relation to the OS). UnRAID doesn't have TRIM pass-through for main (array) drives (..though it does for caching drives), and that's what I chose for his system. Still, the drive space is largely limited in the amount of capacity that is being used with it - from the virtual machines he uses (1 Linux and 1 unregistered Win 10 at the same time.) I've got copies of each - so 4 virtual machines in total, but each machine doesn't have much space allotted to it and UnRAID doesn't seem to be greedily hogging drive space on the array as time goes by (..though it does like to go through memory!),
Last edited by CScottG on Tue Apr 18, 2017 9:37 pm, edited 1 time in total.
 
just brew it!
Gold subscriber
Administrator
Posts: 48246
Joined: Tue Aug 20, 2002 10:51 pm
Location: Somewhere, having a beer

Re: SSD reliability

Tue Apr 18, 2017 9:36 pm

Heh. I just checked, and TRIM isn't enabled for the SSDs in my main desktop. An oversight on my part. But the SSDs are used for OS/system and hosting VMs (home directory is on a mechanical RAID-1 array). So for what the SSDs are being used for, TRIM is probably less important anyway...
If the world isn't making sense to you, you're either drinking too much or not drinking enough.
 
bfg-9000
Gerbil First Class
Posts: 152
Joined: Tue Mar 01, 2016 9:17 pm

Re: SSD reliability

Wed Apr 19, 2017 12:54 am

I have a various X25-M G1 and X25-E that don't support TRIM, that I got for only a few bucks because apparently their performance degraded enough that someone wrote "dead" on a couple with a Sharpie (they are known to slow down below USB 2.0 write speeds if there are no empty blocks to write to). A zero-fill of free space with sdelete fixes them right up.

They are 50nm so should last about forever, with the SLC ones rated 2000TB writes and 100,000 P/E cycles! In comparison my much larger capacity 40nm 3D MLC 850 Pros are "only" rated 150TB writes and 6,000 P/E cycles.
 
Waco
Gold subscriber
Gerbil Jedi
Posts: 1849
Joined: Tue Jan 20, 2009 4:14 pm
Location: Los Alamos, NM

Re: SSD reliability

Wed Apr 19, 2017 12:32 pm

CScottG wrote:
Waco wrote:
I run my 8-way RAID 0 without TRIM and don't miss it a bit.


Not often you hear about using SSD's without TRIM these days. I setup my father's system without TRIM - because I had to (given that he's only using SSD's now, and in relation to the OS). UnRAID doesn't have TRIM pass-through for main (array) drives (..though it does for caching drives), and that's what I chose for his system. Still, the drive space is largely limited in the amount of capacity that is being used with it - from the virtual machines he uses (1 Linux and 1 unregistered Win 10 at the same time.) I've got copies of each - so 4 virtual machines in total, but each machine doesn't have much space allotted to it and UnRAID doesn't seem to be greedily hogging drive space on the array as time goes by (..though it does like to go through memory!),

For write once, read many workloads I find it very hard to care about write performance. The only way to push even remotely modern drives into "bad state" performance is by hammering them mercilessly with random IO...which nobody in the real world does aside from benchmarking.

In normal use I see 2 GB/s+ write speeds even after nearly filling the array which is essentially the rated speed of a single drive * number of drives.
Z170A Gaming Pro Carbon | 6700K @ 4.5 | 16 GB | GTX Titan X | Seasonix Gold 850 | XSPC RX360 | Heatkiller R3 | D5 + RP-452X2 | Cosmos II | 480 + 240 + LSI 9207-8i (128x8) SSDs

Who is online

Users browsing this forum: No registered users and 2 guests