Velocity Reviews - Computer Hardware Reviews

Velocity Reviews > Newsgroups > Computing > Digital Photography > Losing a whole year

Reply
Thread Tools

Losing a whole year

 
 
David Dyer-Bennet
Guest
Posts: n/a
 
      02-10-2012
"David J Taylor" <(E-Mail Removed)> writes:

> "David Dyer-Bennet" <(E-Mail Removed)> wrote in message
> news:(E-Mail Removed)...
> []
>> What's the spec, 10k writes to a cell? So the naive number for a 120GB
>> drive is 1200TB total writes (that assumes perfect wear leveling).
>> Realistic numbers would be considerably lower, but better than half I
>> think. So taking .5 efficiency, that's 1200TB / 2 / 82GB = 7492 days.
>>
>>> Hence my suggestion that before moving to an SSD you consider what I/O
>>> you have, and organise your disks accordingly.

>>
>> Always a good idea, and especially a good idea when you're moving into a
>> realm you're not so familiar with (like when first moving from rotating
>> disk to SSD).

>
> You might be right if every write were just one event, but it's
> not. Unless the file size is /exactly/ a certain size depending on the
> drive, and on /exactly/ the correct boundary, writing will be split
> across more than one block of cells. Don't forget all the directory
> writes as well -
> at create, when my software sets the file modification date, and at erase.


Except for the block alignment issue, the rest of those are included in
an accurate statement of the amount of data written. If they weren't in
your number, then I'd need to know the average file size to estimate
them. If the files were small enough (as in a Usenet server for
example) that meta-data overhead might even dominate. If the files are
large enough, they're pretty much irrelevant. (One block per meta-data
update.)

> I recall read (but please don't ask where) that a life of a few years
> could be expected from and SSD in a read-mostly file server
> application. I won't be recommending my users to go SSD for an
> application writing and deleting tens of thousands of files per day
> with an 80 GB daily throughput. As folk like to store the weather
> data for a few days at least, they will have to go to something like
> one or more 2 TB HDDs in any case.


Well, there's part of your answer; they need amounts of data that would
be expensive on SSD.

> I don't know whether SSDs deserve their reputation for poor
> reliability, or whether its just that we only hear about the failures,
> or whether some folk use them in completely inappropriate ways.


The reputation I hear is that they're much more reliable. You're the
first person I know talking about failures of them.
--
David Dyer-Bennet, http://www.velocityreviews.com/forums/(E-Mail Removed); http://dd-b.net/
Snapshots: http://dd-b.net/dd-b/SnapshotAlbum/data/
Photos: http://dd-b.net/photography/gallery/
Dragaera: http://dragaera.info
 
Reply With Quote
 
 
 
 
nospam
Guest
Posts: n/a
 
      02-10-2012
In article <jh3j39$584$(E-Mail Removed)>, David J Taylor
<(E-Mail Removed)> wrote:

> > What's the spec, 10k writes to a cell? So the naive number for a 120GB
> > drive is 1200TB total writes (that assumes perfect wear leveling).
> > Realistic numbers would be considerably lower, but better than half I
> > think. So taking .5 efficiency, that's 1200TB / 2 / 82GB = 7492 days.
> >
> >> Hence my suggestion that before moving to an SSD you consider what I/O
> >> you have, and organise your disks accordingly.

> >
> > Always a good idea, and especially a good idea when you're moving into a
> > realm you're not so familiar with (like when first moving from rotating
> > disk to SSD).

>
> You might be right if every write were just one event, but it's not.
> Unless the file size is /exactly/ a certain size depending on the drive,
> and on /exactly/ the correct boundary, writing will be split across more
> than one block of cells. Don't forget all the directory writes as well -
> at create, when my software sets the file modification date, and at erase.


7492 days is over 20 years. a lot of what you describe is cached (such
as the directory info), so there won't be as many additional writes as
you suggest, but even assuming there was, an ssd is still very likely
to outlive the product in which it's installed. if it does fail, simply
replace it and restore from a backup, just as you would with a hard
drive failure. ssd only needs to be more reliable than what it's
replacing.

> I recall read (but please don't ask where) that a life of a few years
> could be expected from and SSD in a read-mostly file server application.
> I won't be recommending my users to go SSD for an application writing and
> deleting tens of thousands of files per day with an 80 GB daily
> throughput. As folk like to store the weather data for a few days at
> least, they will have to go to something like one or more 2 TB HDDs in any
> case.


2tb is not cost effective for ssd at this time. that will eventually
change, but for now, hd is better suited for your needs.

> I don't know whether SSDs deserve their reputation for poor reliability,
> or whether its just that we only hear about the failures, or whether some
> folk use them in completely inappropriate ways.


like a lot of things, you mostly hear about the failures. there are
shitloads of hard drive failures too. it's rare someone posts that they
bought a new product and it works great without any problems.
 
Reply With Quote
 
 
 
 
David J Taylor
Guest
Posts: n/a
 
      02-10-2012
"nospam" <(E-Mail Removed)> wrote in message
news:100220121410299079%(E-Mail Removed)...
[]
> like a lot of things, you mostly hear about the failures. there are
> shitloads of hard drive failures too. it's rare someone posts that they
> bought a new product and it works great without any problems.


Yes, that's true. You and David D-B might be interested in this page,
though:

http://www.codinghorror.com/blog/201...ive-scale.html

"[A friend] purchased eight SSDs over the last two years . and all of them
failed."

Cheers,
David

 
Reply With Quote
 
nospam
Guest
Posts: n/a
 
      02-10-2012
In article <jh3uqj$d3h$(E-Mail Removed)>, David J Taylor
<(E-Mail Removed)> wrote:

> > like a lot of things, you mostly hear about the failures. there are
> > shitloads of hard drive failures too. it's rare someone posts that they
> > bought a new product and it works great without any problems.

>
> Yes, that's true. You and David D-B might be interested in this page,
> though:
>
> http://www.codinghorror.com/blog/201...te-drive-scale.
> html
>
> "[A friend] purchased eight SSDs over the last two years . and all of them
> failed."


similar horror stories exist for hard drives. just ask those who had
the 'deathstar' drives from a decade ago or seagates from a few years
ago.
 
Reply With Quote
 
David Dyer-Bennet
Guest
Posts: n/a
 
      02-10-2012
"David J Taylor" <(E-Mail Removed)> writes:

> "nospam" <(E-Mail Removed)> wrote in message
> news:100220121410299079%(E-Mail Removed)...
> []
>> like a lot of things, you mostly hear about the failures. there are
>> shitloads of hard drive failures too. it's rare someone posts that they
>> bought a new product and it works great without any problems.

>
> Yes, that's true. You and David D-B might be interested in this page,
> though:
>
> http://www.codinghorror.com/blog/201...ive-scale.html
>
> "[A friend] purchased eight SSDs over the last two years . and all of
> them failed."


I've only purchased three, but none have failed yet. The oldest is
considerably over two years old.
--
David Dyer-Bennet, (E-Mail Removed); http://dd-b.net/
Snapshots: http://dd-b.net/dd-b/SnapshotAlbum/data/
Photos: http://dd-b.net/photography/gallery/
Dragaera: http://dragaera.info
 
Reply With Quote
 
David J Taylor
Guest
Posts: n/a
 
      02-11-2012
"nospam" <(E-Mail Removed)> wrote in message
news:100220121541487810%(E-Mail Removed)...
[]
> similar horror stories exist for hard drives. just ask those who had
> the 'deathstar' drives from a decade ago or seagates from a few years
> ago.


Yes, I had some of those IBM HDs and, yes, they failed.

David

 
Reply With Quote
 
David J Taylor
Guest
Posts: n/a
 
      02-11-2012
>>
>> http://www.codinghorror.com/blog/201...ive-scale.html
>>
>> "[A friend] purchased eight SSDs over the last two years . and all of
>> them failed."

>
> I've only purchased three, but none have failed yet. The oldest is
> considerably over two years old.
> --
> David Dyer-Bennet, (E-Mail Removed); http://dd-b.net/


I'm hoping that will be my experience as well - I have one just over a
year old. But I have been careful to minimise the number of writes to the
SSD and that is the problem area for them.

Cheers,
David

 
Reply With Quote
 
GMAN
Guest
Posts: n/a
 
      02-11-2012
In article <jh5971$dlk$(E-Mail Removed)>, "David J Taylor" <(E-Mail Removed)> wrote:
>"nospam" <(E-Mail Removed)> wrote in message
>news:100220121541487810%(E-Mail Removed).. .
>[]
>> similar horror stories exist for hard drives. just ask those who had
>> the 'deathstar' drives from a decade ago or seagates from a few years
>> ago.

>


Seagate did free data recovery, or more to the point, they unlocked the locked
drives firmware and updated the firmware then sent back the drives for free.


>Yes, I had some of those IBM HDs and, yes, they failed.
>
>David
>

 
Reply With Quote
 
Robert Coe
Guest
Posts: n/a
 
      02-11-2012
On Tue, 07 Feb 2012 23:50:28 GMT, "MC" <(E-Mail Removed)> wrote:
: (E-Mail Removed) wrote:
:
: > On Sun, 5 Feb 2012 12:46:25 -0800 (PST), otter
: > <(E-Mail Removed)> wrote:
: > > My hard drive crashed, which isn't entirely bad, since I got to
: > > upgrade to a new SSD, and things are spiffy. However, when I went
: > > to recover my files, I found my backup drive hasn't actually been
: > > working - for about a year! Of course, I am pursuing all recovery
: > > options, but I find myself identifying with this song at the moment:
: > >
: > > http://www.youtube.com/watch?v=MwlqymYLCb4
: >
: > It won't help, but I believe important stuff should be backed up a
: > minimum of 3 times. I use 2 hard drives, one of which is stored at a
: > friend's place, and rewriteable Blu Ray disks. Every four years I
: > replace the hard drives. The old ones, still working perfectly, are
: > stored.
:
:
: Agreed. If your photographs are important to you, one backup is just
: asking for trouble.

Can't we allow poor Otter some time to grieve for his lost data before we
start telling him what he should have done? He knows well enough that his
backups weren't in order, and you can bet important money that he won't let
that happen again anytime soon.

I'm lucky not to be walking in Otter's shoes, or worse. I'm responsible for
our backups at work. Three weeks ago, we had a rare double failure of a RAID-1
disk array and lost our backup catalog, which tells where everything lives on
disk and tape. The catalog itself was properly backed up (on a separate RAID
array and on tape), but to use those backups to reconstruct the catalog
requires a magic file that we had allowed to be stored only on the zapped RAID
array. <ulp!!>

Fortunately, we had scheduled an expert on the backup system to come in the
next day to help with a system upgrade. We repurposed his mission, and he and
I spent three days on the phone with the vendor's support group figuring out
how to get our catalog back. We finally managed to rebuild the magic file and
successfully dodged the bullet. But it's gonna be a long time before I throw
stones at somebody else's glass house!

Bob
 
Reply With Quote
 
nospam
Guest
Posts: n/a
 
      02-11-2012
In article <JNpZq.65911$(E-Mail Removed)1.easynews.com>, GMAN
<(E-Mail Removed)> wrote:

> >> similar horror stories exist for hard drives. just ask those who had
> >> the 'deathstar' drives from a decade ago or seagates from a few years
> >> ago.

>
> Seagate did free data recovery, or more to the point, they unlocked the locked
> drives firmware and updated the firmware then sent back the drives for free.


those were the ones with the buggy firmware. that was annoying and very
sloppy on seagate's part, but relatively minor.

i was thinking of seagate's laptop drives from around 5 years ago where
the r/w head had a much higher than normal likelihood of breaking off
from the armature and putting a gouge into the platter, which means the
drive is history.

you *might* be able to have a cleanroom recovery on the non-gouged
portions but that won't be cheap and you will have lost whatever was on
the gouged part.

or, you just grab one of the backups and swap it in.
 
Reply With Quote
 
 
 
Reply

Thread Tools

Posting Rules
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is On
HTML code is Off
Trackbacks are On
Pingbacks are On
Refbacks are Off


Similar Threads
Thread Thread Starter Forum Replies Last Post
Regex losing <br> (different from the earlier topic about losing $1) Jason C Perl Misc 4 06-26-2012 10:29 PM
Losing Drives - Finding Drives - Losing Drives mel@no.spam.com Computer Support 2 09-21-2007 10:16 PM
*WITHOUT* using: ValidateRequest="False" for the whole page (or my whole site).... \A_Michigan_User\ ASP .Net 2 08-21-2006 02:13 PM
Week of year to full Year Otuatail HTML 4 12-08-2003 07:50 PM



Advertisments