ZFS on Linux, new opportunities for MySQL

Database Administration
4 April 12:50PM - 1:40PM @ Ballroom A

Experience level: 
50 minutes conference
These days, on Linux, MySQL typically runs on either xfs or ext4. These filesystems are great performance wise but performance is not the only requirement. For example, backups can generate a high load and take a long time to run. The ZFS filesystem offers an interesting alternative and, although ZFS was available on many OS, ZFS for Linux was only released as GA at the spring of 2013. The talk will first introduce the main concepts surrounding ZFS like pools, filesystems, Copy-on-Write (Cow), ARC, L2ARC, ZIL, TXG, snapshots and clones. From there, we'll demonstrate the installation and configuration of ZFS on Linux and provides some tpcc benchmarks on how ZFS behaves. Guidelines are provided for tuning and best use cases with MyISAM and InnoDB. We will be focusing on the use of ZFS for backups and reporting slaves purposes. We'll walk into the details of setting up a backup solution based on ZFS snapshots and see how it compares with LVM snapshots and Xtrabackup. Finally, we show how to create almost instantaneous live copies of the production data for application testing purposes using ZFS clones.


Principal consultant, Percona
I am a Principal Consultant at Percona, specializing in MySQL High-Availability and scaling solutions. Prior to joining Percona in 2009, I worked as a senior consultant for MySQL AB and Sun Microsystems, assisting customers across North America with NDB Cluster and Heartbeat/DRBD technologies.