[CentOS] best ways to do mysql backup

Jerry Franz jfranz at freerun.com
Sun Aug 15 00:00:05 UTC 2010


On 08/14/2010 03:58 PM, John Hinton wrote:
>
> For uninterrupted delivery of dynamic content from the database... or no
> downtime, replication to a slave is the way to go. This is 'sort of' a
> T-ing effect, except it is to another database. That slave database
> however can be stopped, a mysgldump done to a backup and then restarted,
> at which point the replication restarts and the slave database is
> updated to match the master database. It works really well without huge
> overhead increases.
>
> Google MySQL replication for lots of info about setting it up.
>    

I didn't include this since the OP specified _other_ than tar, 
replication and mysqldump.

But the most efficient and lowest downtime is to combine replication 
with the LVM snapshot and rsync. Since the OP specified he has around 
100 GB of data, rsyncing the snapshotted data directory will be 
substantially more efficient than using mysqldump for transfer to the 
remote system (assuming he doesn't churn most of the contents of the 
database between syncs).

-- 
Benjamin Franz



More information about the CentOS mailing list