page file and SSD life question

Andy Burns wrote:
Paul wrote:

Andy Burns wrote:

Are there scenarios where the memory compressor does useful work?

It's possible the number you're seeing, is the
working set of the memory compressor caching area.

The numbers were just from Task Manager, the "In Use (Compressed)" figure

In this picture, Win10 runs in 256MB
and the Memory Compressor
can be using 30% of the CPU

I never see that process running, it must have a threshold above which
it decides it's not worth running.


The Memory Compressor is "bashful".

It shows in Resource Monitor but not in Task Manager.
I had trouble finding it at first. I couldn't remember
where I'd seen it.

Paul

Paul wrote:

The Memory Compressor is "bashful".
It shows in Resource Monitor but not in Task Manager.

I had quickly looked for it in perfmon like your screenshot, but
overlooked it, so how many other processes does taskmgr hide?

it shows in YAPM as a child of the system process, but it can't see much
info about it.

On 1/20/2017 1:06 AM, T wrote:
Hi All,

What is you guys take of prolonging SSD life by
turning off the page file?

BTW, an SSD's life is long even if you use it for pagefile.sys. Not sure
about bit-torrent downloads though.

https://www.google.com.hk/search?cli...d+life+explain


--
@~@ Remain silent! Drink, Blink, Stretch! Live long and prosper!!
/ v \ Simplicity is Beauty!
/( _ )\ May the Force and farces be with you!
^ ^ (x86_64 Ubuntu 9.10) Linux 2.6.39.3
¤£Â*ɶU! ¤£¶BÄF! ¤£´©¥æ! ¤£¥´¥æ! ¤£¥´§T! ¤£¦Û±þ! ½Ã�¦Ò¼{ºî´© (CSSA):
http://www.swd.gov.hk/tc/index/site_...sub_addressesa

En el artículo , T
escribió:

Yesterday I replaced an Intel SSD (5 year warranty) at a
customer's site. It was only a year and a half old. It
started out withe SMART errors. I told him to leave it off
till I got there so I could recover data from it

Intel SSDs go read-only when they reach end of life, so you have one
chance to copy data off. Power cycling them, however, then renders them
totally unreadable.

You're lucky to get that - lesser SSDs just go offline completely
without warning and total data loss.

Check out SSDLife, a very useful tool:

https://ssd-life.com/

mine (Crucial) have estimated death times of 2023 and 2027 and they're
heavily used in a RAID-0 striped config with the swapfile located on
them. Windows 7.

--
(\_/)
(='.'=) systemd: the Linux version of Windows 10
(")_(")

On Fri, 3 Feb 2017 18:29:16 +0000, Mike Tomlinson
wrote:

En el artículo , T
escribió:

Yesterday I replaced an Intel SSD (5 year warranty) at a
customer's site. It was only a year and a half old. It
started out withe SMART errors. I told him to leave it off
till I got there so I could recover data from it

Intel SSDs go read-only when they reach end of life, so you have one
chance to copy data off. Power cycling them, however, then renders them
totally unreadable.

You're lucky to get that - lesser SSDs just go offline completely
without warning and total data loss.

Check out SSDLife, a very useful tool:

https://ssd-life.com/

mine (Crucial) have estimated death times of 2023 and 2027 and they're
heavily used in a RAID-0 striped config with the swapfile located on
them. Windows 7.

Mine is good until 2025.

Is it now generally accepted that solid state drives are more reliable
than mechanical drives?

On 2/3/2017 10:29 AM, Mike Tomlinson wrote:
En el artículo , T
escribió:

Yesterday I replaced an Intel SSD (5 year warranty) at a
customer's site. It was only a year and a half old. It
started out withe SMART errors. I told him to leave it off
till I got there so I could recover data from it

Intel SSDs go read-only when they reach end of life, so you have one
chance to copy data off. Power cycling them, however, then renders them
totally unreadable.

How does this work?
Is it some measurement of device parameters?
Clock that runs out?
Other?

Turning itself off, total data loss, without warning when it's not
really dead
is a SERIOUS problem for data recovery.
The first step in any debugging process would be
to power cycle the machine.
I can't imagine any user accepting that strategy.

You're lucky to get that - lesser SSDs just go offline completely
without warning and total data loss.

Check out SSDLife, a very useful tool:

https://ssd-life.com/

mine (Crucial) have estimated death times of 2023 and 2027 and they're
heavily used in a RAID-0 striped config with the swapfile located on
them. Windows 7.
2023 is only 7 years from now. I don't think I have ANY spinners
that new. Only drive failure I've had was overheating when the fan quit.

mike wrote:


How does this work?
Is it some measurement of device parameters?
Clock that runs out?
Other?

You should be wary of SSD "wearout detection" and associated policy.

The drive keeps track of gigabytes written per day. So
it does have some metrics it keeps. It does wear leveling,
where it endeavors to write the cells an equal amount.

Say you had TLC chips in an SSD drive. They'd be rated for
3000 writes to each cell. The wear leveling process would
try to maintain an average level of wear across all the pages
of flash. And they count the wear level, which shows as a
"Health indicator" in SMART.

Now, the tricky part, is when the counter hits 3000.

Some brands go read-only - making data recovery easy. Windows
doesn't like to boot on read-only media, so Windows would
probably notice there is a problem if this was C: .

Some other brands, disable all operation. Relying on the
user collection of *backup images* to rescue them. This is
an insane practice, for ordinary human purposes...

It behooves any owner of an SSD, to determine the end-of-life
behavior of the product they bought. And to also check out the
toolkit provided for the product (may involve a download), which
can provide a tool to keep track of what is going on.

In the tiny techie minds of the people who make this
stuff up, maybe they're right. But disabling a device
entirely, when it might actually be fully functional, is
dirty pool. If they're going to entirely disable a device
at detected "end of life", there should be a legal requirement
to have "Make Frequent Backups" printed on the device in
one inch high letters :-)

Paul

On Sat, 04 Feb 2017 00:57:20 -0800, mike wrote:
2023 is only 7 years from now.


Ahem. :-)

--
Stan Brown, Oak Road Systems, Tompkins County, New York, USA
http://BrownMath.com/
http://OakRoadSystems.com/
Shikata ga nai...

En el artículo , mike
escribió:

How does this work?

Google broken, is it?

It's when it runs out of spare flash blocks to remap worn out/faulty
ones.

Turning itself off, total data loss, without warning when it's not
really dead
is a SERIOUS problem for data recovery.

Deal with it, bitch. It's how SSDs work.

The first step in any debugging process would be
to power cycle the machine.
I can't imagine any user accepting that strategy

Don't use an SSD, then.

2023 is only 7 years from now

will you still be using your existing hard drives in 2023?

. I don't think I have ANY spinners
that new

antique dealer, are you?

my oldest drives are 2TB spinning rust from 2007. I use them in RAID5
so that when one fails, the data is still secure.

If you don't do your research before buying an SSD, so are not aware of
the pitfalls, whose fault is that?

--
(\_/)
(='.'=) systemd: the Linux version of Windows 10
(")_(")

En el artículo , Scott
escribió:

Is it now generally accepted that solid state drives are more reliable
than mechanical drives?

No. Just as you can get crappy low-quality mechanical drives or Friday
afternoon drives, so the same applies to SSDs.

The bathtub curve concept also applies to SSDs, perhaps more so.

https://www.quora.com/What-causes-th...ilure-rate-in-
hard-drives

By and large, you get what you pay for.

--
(\_/)
(='.'=) systemd: the Linux version of Windows 10
(")_(")

On 2/4/2017 8:57 AM, Mike Tomlinson wrote:
En el artículo , mike
escribió:

How does this work?

Google broken, is it?

It's when it runs out of spare flash blocks to remap worn out/faulty
ones.

Turning itself off, total data loss, without warning when it's not
really dead
is a SERIOUS problem for data recovery.

Deal with it, bitch. It's how SSDs work.

The first step in any debugging process would be
to power cycle the machine.
I can't imagine any user accepting that strategy

Don't use an SSD, then.

2023 is only 7 years from now

will you still be using your existing hard drives in 2023?

If they still run, youbetcha. This hard drive has been swapped into
updated computers since 2008 or so. S.M.A.R.T. is still all green.
AND it's backed up.

. I don't think I have ANY spinners
that new

antique dealer, are you?

I buy stuff that people like you can't be caught dead with
for pennies on the dollar.


my oldest drives are 2TB spinning rust from 2007. I use them in RAID5
so that when one fails, the data is still secure.

If you don't do your research before buying an SSD, so are not aware of
the pitfalls, whose fault is that?

Hard to do much research at a garage sale. ;-)

My, aren't we full of sunshine this morning.

A working drive that just quits and becomes unrecoverable is UNACCEPTABLE
no matter how snarky you are.

If I loose data because I ignored a warning that keeps popping
up on my screen, that's on me.
If I loose all my data because some counter hit 3000 without
a lot of prior warning, that's UNACCEPTABLE.

I have a SSD sitting in a drawer, and this kind of crap is why
it isn't in a system.

On Sat, 4 Feb 2017 17:03:22 +0000, in alt.comp.os.windows-10, Mike
Tomlinson wrote:

En el artículo , Scott
escribió:

Is it now generally accepted that solid state drives are more reliable
than mechanical drives?

No. Just as you can get crappy low-quality mechanical drives or Friday
afternoon drives, so the same applies to SSDs.

The bathtub curve concept also applies to SSDs, perhaps more so.

https://www.quora.com/What-causes-th...ilure-rate-in-
hard-drives

By and large, you get what you pay for.

I would think the right hand side of the curve is much steeper for SSDs.
They tend to work one day, and just stop outright when they fail, and
eventually they hit the manufacturer's limit.

Any idea of a warning sign that your SSD is about to fail due to wear, or
do you just watch the wear level counter?

With a HDD, there would be gradual failure, and you could catch it in
time by watching SMART. With SSDs, it seems like you need to set a
maximum life window, and replace it at the EOL point you decide upon
whether it is failing or not.

--
Zag

No one ever said on their deathbed, 'Gee, I wish I had
spent more time alone with my computer.' ~Dan(i) Bunten

In message , Zaghadka
writes:
[]
With a HDD, there would be gradual failure, and you could catch it in
time by watching SMART.

Don't _rely_ on that. Mine just stopped suddenly one day.

Though despite that experience, I do _tend_ to agree with you - HDDs _on
the whole_ give warning.

With SSDs, it seems like you need to set a
maximum life window, and replace it at the EOL point you decide upon
whether it is failing or not.

Maybe in time, software (built into the future OS) will take care of
this.
--
J. P. Gilliver. UMRA: 1960/1985 MB++G()AL-IS-Ch++(p)Ar@T+H+Sh0!:`)DNAf

If you believe in telekinesis, raise my right hand

En el artículo , Zaghadka
escribió:

I would think the right hand side of the curve is much steeper for SSDs.

Yes, totally agreed. I did wonder if someone would pick that up.

They tend to work one day, and just stop outright when they fail, and
eventually they hit the manufacturer's limit.

Yes.

Any idea of a warning sign that your SSD is about to fail due to wear, or
do you just watch the wear level counter?

Watch the wear level, and run a tool like SSDLife or Hard Disk Sentinel.
And backup regularly.

With a HDD, there would be gradual failure, and you could catch it in
time by watching SMART.

Agreed.

With SSDs, it seems like you need to set a
maximum life window, and replace it at the EOL point you decide upon
whether it is failing or not.

If your usage case is typical, the SSD is likely to outlast its useful
lifetime.

This article is from 2013, and pretty lengthy, but a very good read for
anyone interested.

http://techreport.com/review/24841/i...ssd-endurance-
experiment

Summary: all the SSDs tested way exceeded their anticipated endurance
limits (translating to a long lifetime under normal day-to-day usage).

--
(\_/)
(='.'=) systemd: the Linux version of Windows 10
(")_(")

On Sat, 04 Feb 2017 07:11:19 -0500, Paul wrote:

====snip====


In the tiny techie minds of the people who make this stuff up, maybe
they're right. But disabling a device entirely, when it might actually
be fully functional, is dirty pool. If they're going to entirely disable
a device at detected "end of life", there should be a legal requirement
to have "Make Frequent Backups" printed on the device in one inch high
letters :-)

That might be a bit tricky to say the least with a 2.5 inch drive. :-)

Some SSD manufacturers (Intel?) may well have added a hari kari action
triggered by reaching the "Guaranteed" 75TBW (or whatever) limit since
that "Test to Destruction" exercise was run.

Whilst there's some legitimacy to blowing the safety fuse of a 30W CFL
on completing a PoH count of 6000 (burning electricity at reduced lamp
efficacy when replacement with a new lamp is guaranteed to provide a
better overall TCO), I don't think such draconian measures are
appropriate with SSDs.

At the very least, it's rather akin to denying the user the benefit of
"Lady Luck"'s munificence in doling out a higher than average service
life to compensate for those statistical cases of premature failure.
Indeed, it's a trick that's guaranteed to shorten the average life of the
product.

The excuse given (and mere excuse it is) for such EoL behaviour is
disabling writes alone leaves the end user with a still readable drive
that can no longer be wiped of user data, hence the complete disabling on
the next power cycle giving the user just one chance to recover data
whilst it's in a read only mode (but only if they're aware that they need
to keep it powered up until a technician can access it with a recovery
boot disk and a backup drive who may need to access the unconnected reset
button pins on the main board to effect the required reboot without
cycling the power - few business grade desktop PCs have a front panel
hardware reset button).

If the issue is one of the end user being unable to securely erase the
SSD because it has exhausted the nand flash cell's erase cycle life, I'm
sure the manufacturers could offer a much better option to the end users
by way of a slotted rotary switch requiring a stout screwdriver and non-
trivial amount of force to invoke the required "self destruct sequence"
mode.

Alternatively, the SSD manufacturer could take a more sanguine view and
not completely disable the SSD, relying instead on the end user's own
good sense to place the data beyond reach *after* retrieving their
precious data, if any, by use of the classic FBH, to batter the damned
SSD to smithereens or roast it slowly over a fire, either method will do
the job quite nicely. :-)

--
Johnny B Good