View Single Post
  #82   Report Post  
Posted to uk.d-i-y,uk.comp.homebuilt
Jaimie Vandenbergh Jaimie Vandenbergh is offline
external usenet poster
 
Posts: 62
Default Building a PC (for those that do) (crossposted)

On Wed, 11 Jul 2012 02:14:00 +0100, John Rumm
wrote:

On 10/07/2012 22:07, Jaimie Vandenbergh wrote:
On Tue, 10 Jul 2012 21:54:25 +0100, John Rumm
wrote:

Add to that the point that your 240G drive will contain significantly
more than 240G of flash. So there are plenty of spare cells available to
patch into the linear addressed space by the wear levelling algorithms
as required.


Ah, knowledge! A question for you if I may: A friend has a
recently-purchased SSD and is trying to interpret the SMART values
he's getting. This is as reported by smartmontools, which supports his
specific SSD (OWC Mercury Electra 6g 480gb). Most of them make sense,
though the 25961524s are a bit opaque.

The main point of interest is the Reallocated_Event_Count = 12; in a
HDD I'd RMA this straight away (surface trouble spreads) but what does
it really mean for an SSD which has overcapacity?


I don't have any specific knowledge on how the smart reporting is used
for that particular drive (and there is some inconsistency between
different brands of SSD - to the point a few don't even report smart
stats at all), so it may be worth asking the manufacturer.


He did, OWC's report back was "Oh, most SMART tools don't really
understand SSDs, it's fine, pay no attention". Which is always the
opposite of reassuring, even if it's true.

As a general point however, SSDs are equipped with significant
overcapacity to enable cells reaching their threshold of write cycles
(or any other failure for that matter) to be remapped. Although on a
grander scale, this is not that different from the way in which spinning
drives also have some set aside capacity for the same purpose.

However the significance of some remapping occurring is different. With
a spinning drive, as you mentioned reallocated sectors are often a
warning of more serious impending drive problems. Too many of them also
impact performance since the add extra thrashing to retrieve sectors
that while consecutive in LBA space, are physically in different places
on a disk. Neither of these are true for SSDs. Correct interpretation of
the stats may let you work out the remaining expected drive life.


The 12 hasn't grown, but he doesn't know if the drive came with them
unfortunately. Interpretation is the problem - there's no references
to interpret from, and you've seen the mfr support comment.

Cheers - Jaimie
--
"What happens if a big asteroid hits Earth? Judging from realistic
simulations involving a sledgehammer and a common laboratory frog,
we can assume it will be pretty bad." - Dave Barry