Effect of partitioning on wear-leveling

[Peter Jeremy]

On 2016-Mar-21 23:45:02 -0700, Hal Murray <hmurray at megapathdsl.net> wrote:
>Has anybody worked out the numbers for something like writing 100 bytes to a 
>log file and flushing things?  I think that turns into 2 page writes, one for 
>the data and one for the directory info.

I assume you mean something like write(2) then fsync(2).  If you're
talking UFS, it's at least 2 synchronous writes (more if you need to
grow the file across a block boundary).  There's also an asynchronous
update to the timestamp in the first superblock.

ZFS is radically different and I'm not sure how many blocks wind up being
written - but around 10 plus 4 per device in the pool seems likely.

-- 
Peter Jeremy
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 949 bytes
Desc: not available
URL: <http://lists.freebsd.org/pipermail/freebsd-arm/attachments/20160322/4dd019c9/attachment.sig>