I fully agree with Morgan’s An argument for not using mysqldump. This post does not come to contradict it, but rather shed a positive light on mysqldump.
I usually prefer an LVM snapshot based backup, or using XtraBackup. And, with databases as large as dozens of GB and above, I consider mysqldump to be a poor alternative. Poor in runtime, poor in overhead while taking the backup.
However once in a while I get to be reminded that mysqldump just works.
As a recent example, I had a server which was killed after an ALTER TABLE statement hanged forever (table already ALTERed, but old scheme never dropped). The old table data still hanged around the file system, but was not recognized by InnoDB. Trying out DISCARD TABLESPACE did not do the job, and eventually file was dropped.
So far, reasonable. InnoDB would complain about some table it never recognized in the first place, but all would work. That is, until backup was concerned. With innobackup or XtraBackup the restore would fail on some internal problem. LVM would work, but would only copy+paste the problem: innobackup would never again be able to be used on this database.
It turned out a 120GB InnoDB compressed data (roughly 250GB uncompressed) would dump in –single-transaction in a matter of 4 hours and would restore in a matter of some 20 hours. A whole lot more than the 3 hours total it would take for an LVM backup for that database. But the data would load well; no missing tablespaces.
I’ve had similar incidents in the past. Not to mention the issue of compressing shared tablespace file.
There’s something about being able to say “I’m not sure how long this is going to take; maybe a day or two. But in the end, we will have problems P1, P2 & P3 resolved“.
I like the clean state you get from a mysqldump restore.
There is yet another thing why I also like mysqldump: you can more easily grep for some data which should or shouldn’t be there which is great for security.
And restore on a completely other platform and/or mysql version also needs mysqldump.
With mysqldump is should also be possible to load data in another RDBMS, but I never thried that.
You can also *pull* a backup from a remote host using mysqldump.
mysqldump has other benefits.
You can and should always dump schema and data separately. This gives you a version of your schema.
mysqldump allows for incremental data recovery, e.g. one table. Even if in one data file you can extract it as it’s text.
You can use mysqldump to modify data during a restore. Ideal for test environments where CLI tools are more effective then DB commands.
mysqldump is the only way to cross OS boundaries.
Hi Shlomi,
I like the clean state as well. I depend on it for moving between major versions – as I do not trust that just using the old data files in a new release will work without issues.
I do not like people to pin me down to saying “what is too large for mysqldump” (because it’s the table structure that really matters). I tell them if they need a number it’s 10G for day-to-day backups. After that you may want to switch.
“I’ve had similar incidents in the past. Not to mention the issue of compressing shared tablespace file.”
A lot of this issue probably comes from actually *running mysqldump*. Using –single-transaction requires old versions of rows need to be kept around while the backup runs, so it can be very dangerous on large systems with a lot of modifications. It also churns all content through the buffer pool, which prior to InnoDB plugin releases can be particularly problematic due to bug #45015. If you stop using mysqldump, you may stop having to do this tablespace compaction 🙂
Compare this to xtrabackup, which uses the end of the backup as a synchronization point – and does not load the data through your InnoDB buffer pool. The way it works is actually very intelligent:
http://www.percona.com/docs/wiki/percona-xtrabackup:xtrabackup:internals#tuning_the_os_buffers
—
Answering some other points:
You do not need to use mysqldump to restore from Windows->Linux or 64-bit to 32-bit as other comments have noted. The only incompatibilities you will have come from case sensitivity, but this can be solved with lower_case_table_names http://dev.mysql.com/doc/refman/5.0/en/identifier-case-sensitivity.html.
You do not need mysqldump to restore a table at a time either. xtrabackup can prepare tables for export (from a MySQL/Percona server), and import them into any Percona server. (Note: a regular MySQL servers can import back to the same server 1 table at a time, but not back to any server).
You do not need mysqldump to pull from another server – xtrabackup can do a streaming backup. Use ssh user@server [cmd] to start the backup and receive it locally.
You do need mysqldump if you want to restore individual rows in an adhoc fashion. This is very common, which is why many people take more than one type of backup. But what I am saying in my original post, is that the primary backup to be used in failure situations should not be mysqldump. That is at least for anyone with just a little data.
mysqldump is needed when migrating from a LSB to MSB architecture.
mysqldump works for all table types. XtraBackup doesn’t work for NDB or custom storage engines.
XtraBackup is the fastest tool for 80% of the backup/restores, for the other 20% there is mysqlbackup.