| View previous topic :: View next topic |
| Author |
Message |
Zarhan
Veteran
Joined: 27 Feb 2004
Posts: 1016
|
 Posted: Mon Jul 12, 2010 7:37 pm Post subject: Ext4 and Discard option (TRIM) |
 |
|
I recently upgraded to X-25 G2 SSD drive from Intel. I've been running ext4. Now, the kernel docs state that
| Code: | discard Controls whether ext4 should issue discard/TRIM
nodiscard(*) commands to the underlying block device when
blocks are freed. This is useful for SSD devices
and sparse/thinly-provisioned LUNs, but it is off
by default until sufficient testing has been done.
|
...sooo, in practice, what does this mean? This is talking about a *filesystem* and I'd sure prefer that I wouldn't get data loss or anything like that, but how "experimental" is this still? What tests have been done and what's the verdict? Should I wait for 2.6.35 or even beyond before setting that option? |
|
| Back to top |
|
 |
ssteinberg
Apprentice
Joined: 09 Jul 2010
Posts: 206
Location: Israel
|
| Posted: Mon Jul 12, 2010 8:05 pm Post subject: |
|
|
Not using an SSD myself yet on home's machine so I can only report what I researched.
TRIM marks an erase block on the SSD as unused (aka free) by the fs. Meaning the next time a write operation takes place on this block the SSD can write without the read->erase->write performance penalty. This is also especially useful for Intel's SSD.
Apparently the TRIM code that comes with the 2.6.34 kernel is still highly unoptimized. Seems like 2.6.35 works well. |
|
| Back to top |
|
|
Zarhan
Veteran
Joined: 27 Feb 2004
Posts: 1016
|
| Posted: Mon Jul 12, 2010 8:27 pm Post subject: |
|
|
| Yeah, I know what TRIM means, but ok, I'll wait for 2.6.35 before turning it on then. |
|
| Back to top |
|
|
|
Kernel & Hardware
