1 |
On Thu, 2010-05-27 at 11:52 +0800, W.Kenworthy wrote: |
2 |
|
3 |
> So something like SMART is the only way for the average Joe to get the |
4 |
> health of a drive. |
5 |
> |
6 |
> Google has lots on this sort of thing |
7 |
|
8 |
That sentence is correct in more than one way! "Google the generic term |
9 |
for any web page found via searching on Google" no doubt has lots on |
10 |
this sort of thing, but so does "Google the company". |
11 |
|
12 |
I remember reading a SlasDot post about the results of their disk |
13 |
monitoring over the last x years, and they use and monitor a lot of |
14 |
disks. |
15 |
|
16 |
*looking* |
17 |
|
18 |
Here's the full study: http://labs.google.com/papers/disk_failures.html |
19 |
|
20 |
>From the abstract: "...we conclude that models based on SMART parameters |
21 |
alone are unlikely to be useful for predicting individual drive |
22 |
failures. Surprisingly, we found that temperature and activity levels |
23 |
were much less correlated with drive failures than previously reported." |
24 |
|
25 |
I recall there were some summaries of this article, but I can't find |
26 |
them right now. |
27 |
|
28 |
An interesting read. Basically, you might not be able to get reliable |
29 |
warnings of impending failures. |
30 |
|
31 |
Keep Good Backups (so say we all) |
32 |
|
33 |
-- |
34 |
Iain Buchanan <iaindb at netspace dot net dot au> |
35 |
|
36 |
The naked truth of it is, I have no shirt. |
37 |
-- William Shakespeare, "Love's Labour's Lost" |