Re: SATA write performance problems (on HP MicroServer)

[Andy Ruhl <acruhl%gmail.com@localhost>]

On Tue, Jan 1, 2013 at 7:00 PM, Michael L. Hitch
<mhitch%lightning.msu.montana.edu@localhost> wrote:
> b)  I wouldn't think so.  In fact, it may possibly reduce the likelyhood of
> a corrupted file system after a crash.  With write caching enabled, you may
> get more of the disk buffers written to the drive cache rather than sitting
> in memory.  Anything in memory will be gone after a crash that is unable to
> flush any buffers.  If the data gets to the drive cache, it should make it
> to the disk eventually.  Now a crash due to a power failure is a different
> story.  Anything in the drive's cache would be lost (but then so would
> anything sitting in memory).  Write caching could also disrupt any ordered
> writes done by the OS, which may get data written to the disk in an order
> that could lead to corruption.

Doesn't this depend on filesystem journaling?

Andy