Have we ever gotten a consensus verdict on whether noatime should be set for 
SSDs or not?

I just received my OWC SSD and installed it today . . .and the speed 
improvement on my 2008 Core 2 Duo Unibody MBP is pretty breathtaking. Following 
advice gleaned from macintouch.com and from Fredrik Poller at 
http://poller.se/2010/08/optimizing-mac-os-x-for-ssd-drives/ I went ahead and 
disabled hibernation (actually  I had already disabled it on my spinning 
drive). Although the article recommends disabling the sudden motion sensor for 
SSDs as it does no good for them . . .disabling it doesn't really provide any 
benefit so I left that alone . . .although I suppose that monitoring it for any 
sudden motion activity might suck up a few cpu cycles but I doubt it would be 
noticeable.

The article further recommends setting noatime to prevent excessive writes to 
the drive. The comments to the article (and related discussions on macintouch) 
argue both for and against the reduction in SSD write-life caused by using 
atime instead of noatime.

Both the pro noatime and pro atime crowd make reasonably convincing arguments 
for their case and claim they're right (imagine that). I can't really see any 
logic flaws in either argument despite reading through them several times. 

So what's the real deal with atime vs noatime? Having the file modified date 
changed just because you modified the file has always seemed kinda hokey to me 
. . .in my view I've always thought that a modification date should reflect the 
date a file was changed and I've never really considered an access a change . . 
.although I can also see the argument that one might want to track accesses as 
well.

Trying to perform an eyeball integration of the various positions . . .what it 
comes down to me is that there is a potential write life lowering if you leave 
atime on . . .but that even granted the theoretical existence of that 
possibility the actual user effect is either significant or not depending on 
whether you agree with the wear leveling portions of the relevant pro/con 
arguments. Beyond that . . .it seems like one should keep atime if you care 
about file accesses being tracked . . .or use noatime if you don't care about 
them and want only actual modifications to the files to be tracked.

So . . .is there any really convincing argument for or against the write life 
argument (at least more convincing than what my research to date has shown me . 
. .which is that it depends on who you ask) . . .and is the effect significant 
enough to use noatime? Conversely  . .is the whole write life argument right up 
there with arguing about whether mailing list software should strip extra 
addresses out or not and you should just use atime or noatime depending on what 
you want tracked?



-----------------------------------------------
There are only three kinds of stress; your basic nuclear stress, cooking 
stress, and A$$hole stress. The key to their relationship is Jello.

neil



_______________________________________________
MacOSX-admin mailing list
[email protected]
http://www.omnigroup.com/mailman/listinfo/macosx-admin

Reply via email to