On 5/2/2013 6:16 AM, Charles Marcus wrote: ...
I've decided to use mdbox for storage (been using maildir), and will enable SIS for attachments.
So, anyone (Stan?) have any suggestions? Should I go with EXT4? Or XFS with just the defaults? Or XFS with one or more tuned parameters?
Appreciate any suggestions (including links to docs dealing with tuning XFS for my mail storage conditions that are written more at the layman level) or comments from anyone experienced using both...
From a filesystem perspective mdbox is little different from maildir as they both exhibit lots of small random IOs. With either one aligning the filesystem to the RAID stripe is problematic as it can create spindle hotspots and increase free space fragmentation. If you're using a vmdk stripe alignment isn't possible anyway as VMware ignores hardware device geometry WRT vmdks.
Although the EXT developers have been working overtime the last few years trying to borrow/steal/duplicate the advanced performance features of XFS, they have a very long way to go. The parallel performance of EXT is far behind as well as file allocation/layout and free space management, to name a few.
My recommendation is to use XFS with the defaults, but add "inode64" to the mount options in /etc/fstab. This enables the modern allocator which clusters files around their parent directory within an allocation group. It's the default allocator in very recent upstream kernels but not in most currently shipping distro kernels. It decreases seek latency between metadata and file operations, and better manages on disk space. In short, XFS will yield superior mail performance to EXT4 in a multiuser environment.
There are currently no mail workload tuning docs in the world of XFS that I'm aware of. I've been intending to write such a doc for the XFS.org FAQ for some time but it hasn't happened yet.
-- Stan