Announcing Percona XtraBackup 1.9.1


Percona is glad to announce the second beta release towards Percona XtraBackup 2.0! We call this version 1.9.1 as we are using the 1.9.x version numbers for betas of 2.0 to clearly differentiate beta releases from the final 2.0 release. We do not recommend using Beta releases in production. Link to BETA version and Release notes.

If you are looking for a stable release for production use, check out our latest stable release (which was 1.6.5 at the time of writing).

This beta adds some really exciting features and bug fixes.

New features include:

  • Backup and restore of LRU dump.
    For servers that support  dump and restore of the XtraDB buffer pool (LRU), namely Percona Server, this will enable a server restored from backup to give excellent performance without a lengthy “warm up” time. If provisioning a slave, this means your new slave will end up with a warmed up buffer pool to mimic that of the master.
  • Streaming incremental backups.
    It is now possible to stream incremental backups, lifting a limitation from previous versions where it was only possible to stream full backups.
  • Parallel compression
    Percona XtraBackup can now use multiple CPUs to perform a compressed backup, significantly reducing the amount of time needed.
  • Parallel compression (‘--compress-threads‘) can be used together with parallel file copying (‘--parallel‘). For example, ‘--parallel=4 --compress --compress-threads=2‘ will create 4 IO threads that will read the data and pipe it to 2 compression threads.
  • New streaming format to work with parallel compression.
    The xbstream format is now supported along with the tar format. The new xbstream format works with parallel compression to (without caching) stream several compressed files at once.
  • tar4ibd functionality is replaced by functionality in the xtrabackup binary itself.
  • The xbstream utility is added
    This utility has a tar like interface to the xbstream format.


  1. says

    We use a hack in tar4ibd which *moves* extracted files instead of copying them.
    Sometimes copying files is not possible due to disk space.
    Could this be an option on the new release?


  2. says

    we’ll have more documentation up for some of the new features “now” (or very soon). We’re also going to improve documentation over the coming weeks, so keep an eye out.

    We’re going to split the documentation for 1.6.x and 2.0 over the next week or so.

  3. Steve Jackson says

    Will innobackex have the possibility to dump structure only for certain tables / entire databases at some point?

    I would like to backup a number of critical database / tables, but some other ones are much less critical, VERY large, and can be rebuilt from other sources

    Therefore, I would like for innobackupex to simply dump the structure for these objects, so I don’t have to do an extra mysqldump -d to cover these…

  4. says

    That’s not a bad idea for a feature – the best ways to have it happen are:
    1) Engage Percona and fund its development
    2) write a patch implementing it
    3) Write up a Blueprint on Launchpad for it and hope somebody implements it.

Leave a Reply

Your email address will not be published. Required fields are marked *