Groups | Search | Server Info | Keyboard shortcuts | Login | Register [http] [https] [nntp] [nntps]

Groups > linux.debian.user > #207081 > unrolled thread

Measuring (or calculating) how many bytes are actually written to disk when I repeatedly save a file

Back to article view | Back to linux.debian.user

Contents

  Measuring (or calculating) how many bytes are actually written to disk when I repeatedly save a file rhkramer@gmail.com - 2019-04-06 19:40 +0200
    Re: Measuring (or calculating) how many bytes are actually written to  disk when I repeatedly save a file "Alexander V. Makartsev" <avbetev@gmail.com> - 2019-04-06 22:40 +0200
    Re: Measuring (or calculating) how many bytes are actually written to disk when I repeatedly save a file Stefan Monnier <monnier@iro.umontreal.ca> - 2019-04-06 23:50 +0200
      Re: Measuring (or calculating) how many bytes are actually written to  disk when I repeatedly save a file Nicholas Geovanis <nickgeovanis@gmail.com> - 2019-04-07 00:40 +0200
      Re: Measuring (or calculating) how many bytes are actually written  to disk when I repeatedly save a file Curt <curty@free.fr> - 2019-04-07 11:50 +0200
        Re: Measuring (or calculating) how many bytes are actually written  to disk when I repeatedly save a file Curt <curty@free.fr> - 2019-04-07 12:00 +0200
        Re: Measuring (or calculating) how many bytes are actually written to disk when I repeatedly save a file rhkramer@gmail.com - 2019-04-07 15:10 +0200
      Re: Measuring (or calculating) how many bytes are actually written to disk when I repeatedly save a file rhkramer@gmail.com - 2019-04-07 14:20 +0200
        Re: Measuring (or calculating) how many bytes are actually written  to disk when I repeatedly save a file Erik Christiansen <dvalin@internode.on.net> - 2019-04-07 15:00 +0200
        Re: Measuring (or calculating) how many bytes are actually written to disk when I repeatedly save a file Stefan Monnier <monnier@iro.umontreal.ca> - 2019-04-07 18:30 +0200
    Re: Measuring (or calculating) how many bytes are actually written  to disk when I repeatedly save a file Andy Smith <andy@strugglers.net> - 2019-04-07 03:30 +0200
      Re: Measuring (or calculating) how many bytes are actually written to disk when I repeatedly save a file rhkramer@gmail.com - 2019-04-07 15:00 +0200
    Re: [OFF-LIST] Re: Measuring (or calculating) how many bytes are actually written to disk when I repeatedly save a file rhkramer@gmail.com - 2019-04-07 14:00 +0200
    Re: Measuring (or calculating) how many bytes are actually written  to disk when I repeatedly save a file Carles Pina i Estany <carles@pina.cat> - 2019-04-07 22:30 +0200
      Re: Measuring (or calculating) how many bytes are actually written  to disk when I repeatedly save a file Reco <recoverym4n@enotuniq.net> - 2019-04-07 22:30 +0200
        Re: Measuring (or calculating) how many bytes are actually written  to disk when I repeatedly save a file Curt <curty@free.fr> - 2019-04-08 09:50 +0200
          Re: Measuring (or calculating) how many bytes are actually written to disk when I repeatedly save a file rhkramer@gmail.com - 2019-04-08 14:50 +0200
            Re: Measuring (or calculating) how many bytes are actually written to disk when I repeatedly save a file rhkramer@gmail.com - 2019-04-08 15:40 +0200
            Re: Measuring (or calculating) how many bytes are actually written  to disk when I repeatedly save a file Curt <curty@free.fr> - 2019-04-08 15:40 +0200
              Re: Measuring (or calculating) how many bytes are actually written to disk when I repeatedly save a file rhkramer@gmail.com - 2019-04-08 15:50 +0200
                Re: Measuring (or calculating) how many bytes are actually written  to disk when I repeatedly save a file Curt <curty@free.fr> - 2019-04-08 16:20 +0200
                  Re: Measuring (or calculating) how many bytes are actually written to disk when I repeatedly save a file rhkramer@gmail.com - 2019-04-08 17:00 +0200
                    Re: Measuring (or calculating) how many bytes are actually written to disk when I repeatedly save a file Gene Heskett <gheskett@shentel.net> - 2019-04-08 18:20 +0200
                    Re: Measuring (or calculating) how many bytes are actually written  to disk when I repeatedly save a file Curt <curty@free.fr> - 2019-04-08 19:00 +0200
                    Re: Measuring (or calculating) how many bytes are actually written to  disk when I repeatedly save a file "Alexander V. Makartsev" <avbetev@gmail.com> - 2019-04-08 19:20 +0200
              Re: Measuring (or calculating) how many bytes are actually written  to disk when I repeatedly save a file Andy Smith <andy@strugglers.net> - 2019-04-12 11:20 +0200
                Re: Measuring (or calculating) how many bytes are actually written  to disk when I repeatedly save a file Curt <curty@free.fr> - 2019-04-12 14:10 +0200
                  Re: Measuring (or calculating) how many bytes are actually written to disk when I repeatedly save a file rhkramer@gmail.com - 2019-04-12 19:20 +0200
                    Re: Measuring (or calculating) how many bytes are actually written  to disk when I repeatedly save a file Michael Stone <mstone@debian.org> - 2019-04-12 20:50 +0200
                      Re: Measuring (or calculating) how many bytes are actually written to  disk when I repeatedly save a file "Alexander V. Makartsev" <avbetev@gmail.com> - 2019-04-12 22:50 +0200
            Re: Measuring (or calculating) how many bytes are actually written to  disk when I repeatedly save a file "Alexander V. Makartsev" <avbetev@gmail.com> - 2019-04-08 15:50 +0200
              Re: Measuring (or calculating) how many bytes are actually written to disk when I repeatedly save a file rhkramer@gmail.com - 2019-04-08 18:00 +0200
        Re: Measuring (or calculating) how many bytes are actually written to disk when I repeatedly save a file rhkramer@gmail.com - 2019-04-08 14:40 +0200
          Re: Measuring (or calculating) how many bytes are actually written to  disk when I repeatedly save a file David <bouncingcats@gmail.com> - 2019-04-08 16:40 +0200

Page 2 of 2 — ← Prev page 1 [2]

#207135 — Re: Measuring (or calculating) how many bytes are actually written to disk when I repeatedly save a file

On 2019-04-08, rhkramer@gmail.com <rhkramer@gmail.com> wrote:
>
> And someone would (or should) ask what does "frequently" mean, and that is 
> what I am trying to quantify.  

Sure. But below a certain level of granularity it becomes an exercise
for which the benefits remain to be established.  Large files and
frequent writes to disk are probably not ideal for an SSD. You can read
that anywhere. What else can anyone say? What are you trying to do?
Measure the daily byte-count of your writes and compare that to some
hypothetical standard or threshold above which it would be suggested to
employ an HDD rather than a SSD? 

[toc] | [prev] | [next] | [standalone]

#207138

On Monday, April 08, 2019 10:18:28 AM Curt wrote:
> On 2019-04-08, rhkramer@gmail.com <rhkramer@gmail.com> wrote:
> > And someone would (or should) ask what does "frequently" mean, and that
> > is what I am trying to quantify.
> 
> Sure. But below a certain level of granularity it becomes an exercise
> for which the benefits remain to be established.  Large files and
> frequent writes to disk are probably not ideal for an SSD. You can read
> that anywhere. What else can anyone say? 

I've seen that implied, but not explicitly stated, nor with "quantification" 
(i.e., supporting evidence).

> What are you trying to do?
> Measure the daily byte-count of your writes and compare that to some
> hypothetical standard or threshold above which it would be suggested to
> employ an HDD rather than a SSD?

As mentioned in another post, I am starting to fear for the reilability of an 
HDD (DOAs, early failures, unwilingness of the vendor / manufacturer to 
provide a warranty), and, therefore, I am trying to determine if an SSD could 
be a better choice.

(Someday, I expect it will be -- is that day here?)

[toc] | [prev] | [next] | [standalone]

#207142

On Monday 08 April 2019 10:56:33 rhkramer@gmail.com wrote:

> On Monday, April 08, 2019 10:18:28 AM Curt wrote:
> > On 2019-04-08, rhkramer@gmail.com <rhkramer@gmail.com> wrote:
> > > And someone would (or should) ask what does "frequently" mean, and
> > > that is what I am trying to quantify.
> >
> > Sure. But below a certain level of granularity it becomes an
> > exercise for which the benefits remain to be established.  Large
> > files and frequent writes to disk are probably not ideal for an SSD.
> > You can read that anywhere. What else can anyone say?
>
> I've seen that implied, but not explicitly stated, nor with
> "quantification" (i.e., supporting evidence).
>
> > What are you trying to do?
> > Measure the daily byte-count of your writes and compare that to some
> > hypothetical standard or threshold above which it would be suggested
> > to employ an HDD rather than a SSD?
>
> As mentioned in another post, I am starting to fear for the
> reilability of an HDD (DOAs, early failures, unwilingness of the
> vendor / manufacturer to provide a warranty), and, therefore, I am
> trying to determine if an SSD could be a better choice.
>
> (Someday, I expect it will be -- is that day here?)

I don't think so, so where I do use them (the speed is nice!!!), I use 
them to maybe 20% of capacity so the drive has plenty of room to 
reassign a questionable spot. So as rust fails, I'm installing 40 to 60 
GB SSD's in my machine tools as the OS and files don't exceed around 5Gb 
even after years of carving metal.  And I have been doing so for over a 
year with zero problems.

OTOH, I have a now ancient seacrate 1Tb with 80,000+ spinning hours on 
it, had 25 reallocated sectors when I updated the firmware in it, 
gaining 30 mb/second in speeds both ways when it was about a month old. 
Its had its total contents rewritten at least 500 times as its been a 
virtual tape library for amanda since it was bought.  But when it hit 
90% full, I figured it was time for a new 2Tb. Currently saving the 
virtual tapes of a 5 machine network every night.  Its a good sata-II 
drive yet, still after all that time is reporting the same 25 
re-allocated sectors, and altho its out of a job, its still spinning 
because if shut down for several days, it will probably be hard to start 
because of stiction.

I have a pair of 1Gb seagates attached to a legacy computer in the 
basement, that probably have in excess of 200,000 hours on them. If shut 
down for 2 weeks, will need a gentle sideways tap with a rubber hammer 
on a front corner to start them, but neither has dropped a single bit 
since I took them out of a dying amiga nearly 15 years ago.
Spinning rust can be very dependable.

Cheers, Gene Heskett
-- 
"There are four boxes to be used in defense of liberty:
 soap, ballot, jury, and ammo. Please use in that order."
-Ed Howdershelt (Author)
Genes Web page <http://geneslinuxbox.net:6309/gene>

[toc] | [prev] | [next] | [standalone]

#207143 — Re: Measuring (or calculating) how many bytes are actually written to disk when I repeatedly save a file

On 2019-04-08, rhkramer@gmail.com <rhkramer@gmail.com> wrote:
>
> As mentioned in another post, I am starting to fear for the reilability of an 
> HDD (DOAs, early failures, unwilingness of the vendor / manufacturer to 
> provide a warranty), and, therefore, I am trying to determine if an SSD could 
> be a better choice.
>
> (Someday, I expect it will be -- is that day here?)
>
>

Here's one study for you (where number of writes and enterprise-grade
drives--mentioned by someone in this thread I believe--are found not to be
determining factors for longevity or reliability):

https://www.zdnet.com/article/ssd-reliability-in-the-real-world-googles-experience/

http://0b4af6cdc2f0c5998459-c0245c5c937c5dedcca3f1764ecc9b2f.r43.cf2.rackcdn.com/23105-fast16-papers-schroeder.pdf

 Two standout conclusions from the study. First, that MLC drives are as
 reliable as the more costly SLC "enteprise" drives. This mirrors hard drive
 experience, where consumer SATA drives have been found to be as reliable as
 expensive SAS and Fibre Channel drives.

 One of the major reasons that "enterprise" SSDs are more expensive is due to
 greater over-provisioning. SSDs are over-provisioned for two main reasons: to
 allow for ample bad block replacement caused by flash wearout; and, to ensure
 that garbage collection does not cause write slowdowns.

 The paper's second major conclusion, that age, not use, correlates with
 increasing error rates, means that over-provisioning for fear of flash wearout
 is not needed. None of the drives in the study came anywhere near their write
 limits, even the 3,000 writes specified for the MLC drives.

 But it isn't all good news. SSD UBER rates are higher than disk rates, which
 means that backing up SSDs is even more important than it is with disks. The
 SSD is less likely to fail during its normal life, but more likely to lose
 data.

[toc] | [prev] | [next] | [standalone]

#207145 — Re: Measuring (or calculating) how many bytes are actually written to disk when I repeatedly save a file

[Multipart message — attachments visible in raw view] — view raw

On 08.04.2019 19:56, rhkramer@gmail.com wrote:
> On Monday, April 08, 2019 10:18:28 AM Curt wrote:
>> On 2019-04-08, rhkramer@gmail.com <rhkramer@gmail.com> wrote:
>>> And someone would (or should) ask what does "frequently" mean, and that
>>> is what I am trying to quantify.
>> Sure. But below a certain level of granularity it becomes an exercise
>> for which the benefits remain to be established.  Large files and
>> frequent writes to disk are probably not ideal for an SSD. You can read
>> that anywhere. What else can anyone say? 
> I've seen that implied, but not explicitly stated, nor with "quantification" 
> (i.e., supporting evidence).
>
>> What are you trying to do?
>> Measure the daily byte-count of your writes and compare that to some
>> hypothetical standard or threshold above which it would be suggested to
>> employ an HDD rather than a SSD?
> As mentioned in another post, I am starting to fear for the reilability of an 
> HDD (DOAs, early failures, unwilingness of the vendor / manufacturer to 
> provide a warranty), and, therefore, I am trying to determine if an SSD could 
> be a better choice.
>
> (Someday, I expect it will be -- is that day here?)
>
>
Don't know if it is appropriate, but you should begin to treat any media
(HDD, SSD, etc) as consumables.
You can't expect some device will work for any amount of time and won't
fail. They all fail. Even tanks fail.
Any media could fail for sooo many different reasons and nobody can
predict them all. You have to be always prepared for the moment when
that happens, basically
by asking yourself this question: "What will I do if my current disk
will fail right at this moment and there will be no possibility to
recover any data from it?"
If your answer is: "I will replace it and restore my data from backup on
another device and probably loose a day worth of work." Then you're fine.
But if your answer is: "This is not happening..This is not
happening..This is not happening.." Then it's a huge problem, and it has
nothing to do with device's reliability or type.
So stop fearing and perform regular backups.

That said, personally I can't imagine a PC without a SSD or NVMe drive
nowadays. Even more if it is a laptop.
SSDs are relatively cheap, super fast and became more reliable, than
they were a few years ago. Firmwares got better, controller ICs got better.
HDDs has their uses too and they mostly used for storage capacity, in
workstations, NAS devices and servers with RAID for redundancy.
If your fear is about warranty and RMA then buy devices from well known
brands and from your local PC hardware store. Everything else is
mitigated by backups.

-- 
With kindest regards, Alexander.

⢀⣴⠾⠻⢶⣦⠀ 
⣾⠁⢠⠒⠀⣿⡁ Debian - The universal operating system
⢿⡄⠘⠷⠚⠋⠀ https://www.debian.org
⠈⠳⣄⠀⠀⠀⠀ 

[toc] | [prev] | [next] | [standalone]

#207365 — Re: Measuring (or calculating) how many bytes are actually written to disk when I repeatedly save a file

Hello,

On Mon, Apr 08, 2019 at 01:32:48PM -0000, Curt wrote:
> How about:
> 
>  Subject: SSD for frequent edits of large text files?

It's really hard for me to imagine any form of human editing of a
text file that could wear out a modern SSD. Natural language text
files just aren't that big, and human fingers and brains just don't
operate that fast.

To wear these things out you need multiple users, vast numbers of
small writes (because the erase size of an SSD is typically 1MiB or
more, so at minimum it writes that every time), things of that
nature. One person revising their memoirs for example is not going
to hit it, even if they are as loquacious and capricious as Richard
Owlett avoiding giving a direct answer to a reasonable question.

Honestly my advice to the OP as suggested what seems like many days
ago remains: just take a measure, do a day or two of work, take
another measure, check the difference in byte count and extrapolate
from there. I'd be amazed if you didn't end up with multiple decades
of write headroom.

Too much has been written here on this subject without actual
testing of the realities. We can debate forever how many angels can
dance on the head of a pin, but in this case both the pin and the
angels are extremely easy to quantify as they come with spec sheets
and SMART attributes. Perhaps it is an attempt to exhaust our
brains' collective write endurance.

Cheers,
Andy

-- 
https://bitfolk.com/ -- No-nonsense VPS hosting

[toc] | [prev] | [next] | [standalone]

#207373 — Re: Measuring (or calculating) how many bytes are actually written to disk when I repeatedly save a file

On 2019-04-12, Andy Smith <andy@strugglers.net> wrote:
>
> Honestly my advice to the OP as suggested what seems like many days
> ago remains: just take a measure, do a day or two of work, take
> another measure, check the difference in byte count and extrapolate
> from there. I'd be amazed if you didn't end up with multiple decades
> of write headroom.

No need to even measure.

https://en.wikipedia.org/wiki/Solid-state_drive#SSD_reliability_and_failure_modes

 Device age, measured by days in use, is the main factor in SSD
 reliability, and not amount of data read or written, which are measured
 by TBW or DWPD.

http://0b4af6cdc2f0c5998459-c0245c5c937c5dedcca3f1764ecc9b2f.r43.cf2.rackcdn.com/23105-fast16-papers-schroeder.pdf

The goose has been wild from the beginning, and the air, hot.

[toc] | [prev] | [next] | [standalone]

#207391

On Friday, April 12, 2019 08:07:07 AM Curt wrote:
> On 2019-04-12, Andy Smith <andy@strugglers.net> wrote:
> > Honestly my advice to the OP as suggested what seems like many days
> > ago remains: just take a measure, do a day or two of work, take
> > another measure, check the difference in byte count and extrapolate
> > from there. I'd be amazed if you didn't end up with multiple decades
> > of write headroom.
> 
> No need to even measure.
> 
> https://en.wikipedia.org/wiki/Solid-state_drive#SSD_reliability_and_failure
> _modes
> 
>  Device age, measured by days in use, is the main factor in SSD
>  reliability, and not amount of data read or written, which are measured
>  by TBW or DWPD.
> 
> http://0b4af6cdc2f0c5998459-c0245c5c937c5dedcca3f1764ecc9b2f.r43.cf2.rackcd
> n.com/23105-fast16-papers-schroeder.pdf

I read (or skimmed) these the first time you posted them (well, the first one 
might have been a different URL with similar information).  

Unless I missed something, neither one of them addresses 3-D NAND drives, 
which presumably would not do as well.  (Well, they might exceed their 
published TBW figures, but those figures are, iirc, lower than those for SLC or 
MLC).

It looks like I am writing something like 20 GB per day, or something like 8 
TB per year (due to my quick trigger on <ctrl>s).  I plan to test over a 
longer period of time -- right now I am doing a little less writing due to tax 
season.

[toc] | [prev] | [next] | [standalone]

#207395 — Re: Measuring (or calculating) how many bytes are actually written to disk when I repeatedly save a file

On Fri, Apr 12, 2019 at 01:13:49PM -0400, rhkramer@gmail.com wrote:
>It looks like I am writing something like 20 GB per day

That's basically nothing for a reasonably sized modern SSD.

[toc] | [prev] | [next] | [standalone]

#207401 — Re: Measuring (or calculating) how many bytes are actually written to disk when I repeatedly save a file

[Multipart message — attachments visible in raw view] — view raw

On 12.04.2019 23:47, Michael Stone wrote:
> On Fri, Apr 12, 2019 at 01:13:49PM -0400, rhkramer@gmail.com wrote:
>> It looks like I am writing something like 20 GB per day
>
> That's basically nothing for a reasonably sized modern SSD.
>
Still the SSD market is having a steady decline in terms of
quality\longevity of NAND based devices.
SSD manufacturers quickly discovered that drives they sell are too good
and last too long.
So now we have TLC-based and QLC-based drives for almost all market
segments, which sold for almost the same price as MLC-based drives were
sold in the past.
Ex. Kingston HyperX FURY [SHFS37A/120G] was produced in 2014 and was
MLC(2bit)-based. It had 3 year warranty and was rated for 354 TBW or
2,69 DWPD.
HyperX Fury RGB [SHFR200/240G] (2019) is now 3D TLC(3bit)-based. It has
same 3 year limited warranty and advertised as 120 TBW or 0,46 DWPD.

Worst offender of this "modern SSD fraud" is Samsung, because he still
says 3D-NAND-V-whatever MLC in advertisements when in fact those drives
are TLC, because TLC NAND is 3bit per cell.
Only recently, Samsung began to add (3bit) to that "3D V-NAND MLC"
phrasing in specifications.
Other manufacturers parroting the same marketing trick and getting away
with it.
And as a result, MLC-based drives almost vanished from
consumer\enthusiast markets because people are happy with TLC and know
no better.


-- 
With kindest regards, Alexander.

⢀⣴⠾⠻⢶⣦⠀ 
⣾⠁⢠⠒⠀⣿⡁ Debian - The universal operating system
⢿⡄⠘⠷⠚⠋⠀ https://www.debian.org
⠈⠳⣄⠀⠀⠀⠀ 

[toc] | [prev] | [next] | [standalone]

#207132 — Re: Measuring (or calculating) how many bytes are actually written to disk when I repeatedly save a file

[Multipart message — attachments visible in raw view] — view raw

On 08.04.2019 17:39, rhkramer@gmail.com wrote:
> On Monday, April 08, 2019 03:40:54 AM Curt wrote:
>> Maybe an SSD is not the most appropriate
>> storage device for frequent editing of large files.
> That is what I'm trying to decide / determine.

There are NVMe drives and SSDs intended to be used in servers with high
workloads like cache storage. These server grade drives must be rated
for at least 3 DWPD (Drive Writes Per Day) or more.
This means, if you have a 500GB SSD rated for 3 DWPD and it has 5 year
warranty, then you can safely write 1,5TB to it each day for next 5 years.
That is 3 * 500 * ( 365 * 5 ) / 1000 = 2737TB in total. (TBW)

You can convert TBW to DWPD if you want.
Let's say we have 240GB SSD rated for 768 TBW and it has 5 year warranty.
That will be ( 768 * 1000 ) / ( 365 * 5 ) = 420GB could be written per
day, or 420 / 240 = 1,75 DWPD

Let's say we have 256GB SSD rated for 300TBW and it has 5 year warranty.
That will be ( 300 * 1000 ) / 1825 = 164GB could be written per day, or
164 / 256 =  0,64 DWPD

IMO, an average consumer grade SSD (preferably MLC NAND based, 240GB+, 5
year warranty) should be rated at least 1 DWPD to be worth buying, so it
could be used (without paying constant attention to it,
implementing various tricks and restrictions to its workload, etc) for a
very long period of time, extending far beyond its warranty period, if
your workload is lower than 1 DWPD.
However, in reality it is so much trouble just to find suitable device.
This involves browsing through terribly designed manufacturer websites
with dark marketing patterns [1] and a pile of specification datasheet
files.
Also keeping in mind how SSD manufacturers don't like to talk about this
inconvenient topic for them, trying to hide TBW DWPD ratings by using
MTBF ratings instead.



[1] https://darkpatterns.org/

-- 
With kindest regards, Alexander.

⢀⣴⠾⠻⢶⣦⠀ 
⣾⠁⢠⠒⠀⣿⡁ Debian - The universal operating system
⢿⡄⠘⠷⠚⠋⠀ https://www.debian.org
⠈⠳⣄⠀⠀⠀⠀ 

[toc] | [prev] | [next] | [standalone]

#207141

[Multipart message — attachments visible in raw view] — view raw

On Monday, April 08, 2019 09:41:10 AM Alexander V. Makartsev wrote:
> There are NVMe drives and SSDs intended to be used in servers with high
> workloads like cache storage. These server grade drives must be rated
> for at least 3 DWPD (Drive Writes Per Day) or more.

Thanks!  I really hadn't encountered the acronym NVMe before.

Looking at some (e.g., by description "SAMSUNG 970 PRO M.2 2280 512GB PCIe 
Gen3. X4, NVMe 1.3 64L V-NAND 2-bit MLC Internal Solid State 
Drive (SSD) MZ-V7P512BW") seem to be the older style of SSD 
technology (V-NAND 2-bit MLC) which might be a little more 
reliable (though more expensive), than 

some others which use the 3-D NAND technology, which presumably has about the 
same longevity / reliability as devices labeled as SSDs which use the 3-D NAND 
technology -- so I guess I have to be careful in how I shop (which is always 
the case, anyway).

Two links, both in the same ballpark pricewise:

   * [[https://www.newegg.com/Product/Product.aspx?Item=N82E16820147693]
[SAMSUNG 970 PRO M.2 2280 512GB PCIe Gen3. X4, NVMe 1.3 64L V-NAND 2-bit MLC 
Internal Solid State Drive (SSD) MZ-V7P512BW]]

   * [[https://www.amazon.com/BLACK-SN750-NVMe-Internal-
Gaming/dp/B07M64QXMN?tag=pcworld02-20&psc=1&ascsubtag=US-001-2899351-000-1442373-
web-20][WD BLACK SN750 1TB NVMe Internal Gaming SSD - Gen3 PCIe, M.2 2280, 3D 
NAND - WDS100T3X0C]]

And thanks for the explanation / clarification on some of the terminology:

> This means, if you have a 500GB SSD rated for 3 DWPD and it has 5 year
> warranty, then you can safely write 1,5TB to it each day for next 5 years.
> That is 3 * 500 * ( 365 * 5 ) / 1000 = 2737TB in total. (TBW)
> 
> You can convert TBW to DWPD if you want.
> Let's say we have 240GB SSD rated for 768 TBW and it has 5 year warranty.
> That will be ( 768 * 1000 ) / ( 365 * 5 ) = 420GB could be written per
> day, or 420 / 240 = 1,75 DWPD
> 
> Let's say we have 256GB SSD rated for 300TBW and it has 5 year warranty.
> That will be ( 300 * 1000 ) / 1825 = 164GB could be written per day, or
> 164 / 256 =  0,64 DWPD
> 
> IMO, an average consumer grade SSD (preferably MLC NAND based, 240GB+, 5
> year warranty) should be rated at least 1 DWPD to be worth buying, so it
> could be used (without paying constant attention to it,
> implementing various tricks and restrictions to its workload, etc) for a
> very long period of time, extending far beyond its warranty period, if
> your workload is lower than 1 DWPD.
> However, in reality it is so much trouble just to find suitable device.
> This involves browsing through terribly designed manufacturer websites
> with dark marketing patterns [1] and a pile of specification datasheet
> files.
> Also keeping in mind how SSD manufacturers don't like to talk about this
> inconvenient topic for them, trying to hide TBW DWPD ratings by using
> MTBF ratings instead.
> 
> 
> 
> [1] https://darkpatterns.org/

[toc] | [prev] | [next] | [standalone]

#207121

On Sunday, April 07, 2019 04:22:41 PM Reco wrote:
> On Sun, Apr 07, 2019 at 10:10:58PM +0200, Carles Pina i Estany wrote:

> > In my SSDs I have:
> > /sys/fs/ext4/dm-0/lifetime_write_kbytes
> > 
> > I'm not sure if this is specific for SSD?
> 
> No, it's not. It's filesystem-specific though.
> Meaning - you have to use ext4 to see this attribute, but the device
> where the ext4 filesystem resides does not matter.

Well, to clarify, if you have multiple ext4  filesystems, does that represent 
the sum of lifetime_write_kbytes of all of those filesystems?  

(And,iiuc, any ext2 (or others) are not included in that sum.)

[toc] | [prev] | [next] | [standalone]

#207136 — Re: Measuring (or calculating) how many bytes are actually written to disk when I repeatedly save a file

On Mon, 8 Apr 2019 at 22:38, <rhkramer@gmail.com> wrote:
> On Sunday, April 07, 2019 04:22:41 PM Reco wrote:
> > On Sun, Apr 07, 2019 at 10:10:58PM +0200, Carles Pina i Estany wrote:
>
> > > In my SSDs I have:
> > > /sys/fs/ext4/dm-0/lifetime_write_kbytes
> > >
> > > I'm not sure if this is specific for SSD?
> >
> > No, it's not. It's filesystem-specific though.
> > Meaning - you have to use ext4 to see this attribute, but the device
> > where the ext4 filesystem resides does not matter.
>
> Well, to clarify, if you have multiple ext4  filesystems, does that represent
> the sum of lifetime_write_kbytes of all of those filesystems?

No, the device is in the path, after the filesystem type.
The device above is dm-0. Here's an example from this machine,
for device sda10:
$ cat /sys/fs/ext4/sda10/lifetime_write_kbytes
8589077

A quick search found this (it's not recent) which might give you
some things to try:
https://serverfault.com/questions/238033/measuring-total-bytes-written-under-linux

[toc] | [prev] | [standalone]

Page 2 of 2 — ← Prev page 1 [2]

Back to top | Article view | linux.debian.user

csiph-web