On Oct 15, 2009, at 1:19 PM, Javier Conde wrote:
Hello,
I've seen in the "what's new" of Solaris 10 update 8 just released
that ZFS now includes the "primarycache" and "secondarycache"
properties.
Is this the "equivalent" of the UFS directio?
No. UFS directio does 3 things:
1. unbuffered I/O
2. allow concurrent writers (no single-writer lock)
3. provide an improved async I/O code path
For #1, the ZFS equivalent is primarycache=none, but you might find that
primarycache=metadata is more appropriate, depending on your workload.
For #2 & #3, ZFS did not have these limitations from the start.
Does it have a similar behavior?
As you can see, this question doesn't really make sense because the
definition of "directio" is not consistent across the many
implementations
calling themselves "*directio*" :-)
I'm thinking about having a database on ZFS with this option, and
Oracle recommends to have directio when working on top of a file
system.
As always, YMMV. But in general, I think primarycache=metadata is a
reasonable default for Oracle workloads on ZFS file systems or volumes
when there is no L2ARC (cache) device. Tests clearly show that the
L2ARC can benefit Oracle database workloads and represent a new way
to look at the price/performance analysis for databases.
http://blogs.sun.com/fishworks/resource/oow09_ss7000_oracle.pdf
-- richard
_______________________________________________
zfs-discuss mailing list
zfs-discuss@opensolaris.org
http://mail.opensolaris.org/mailman/listinfo/zfs-discuss
