XtraDB Features

The list of main XtraDB features that are not available in InnoDB-plugin

• Improved buffer_pool scalability | innodb_split_buf_pool_mutex
  • This change takes advantage of multiple threads and adaptive flushing to achieve better write performance, particularly with FusionIO SSD cards. We have seen about 30% to 50% more writes with FusionIO cards after making this change. The best improvement comes when using at least 4 and up to 16 threads simultaneously. For more information see:
  • http://www.mysqlperformanceblog.com/2010/01/13/innodb-innodb-plugin-vs-xtradb-on-fast-storage/
• Fast recovery | innodb_recovery_patches
  • http://www.mysqlperformanceblog.com/2009/07/07/improving-innodb-recovery-time/
• Improved IO path | innodb_io
  • http://www.mysqlperformanceblog.com/2010/01/13/innodb-innodb-plugin-vs-xtradb-on-fast-storage/
• Improved rollback segment scalability | innodb_extra_rseg
  • http://www.mysqlperformanceblog.com/2009/01/18/partial-fix-of-innodb-scalability-rollback-segments/
• Separate purge thread | innodb_purge_thread
• Limited size of data dictionary in memory | innodb_dict_size_limit
  • http://www.mysqlperformanceblog.com/2009/02/11/limiting-innodb-data-dictionary/
• Increased number of concurrent write transactions (undo slots) ( up to 4000 ) | innodb_expand_undo_slots
• Fast checksums ( in release process )
• Support of different pagesizes ( 4K, 8K, 16K) ( in release process )

• Show content of buffer_pool | innodb_buffer_pool_pages
  • http://www.mysqlperformanceblog.com/2010/03/26/tables-fit-buffer-poo/
• Import / export of dedicated tables | innodb_expand_import
  • http://www.mysqlperformanceblog.com/2009/07/31/copying-innodb-tables-between-servers/
  • http://www.mysqlperformanceblog.com/2009/06/08/impossible-possible-moving-innodb-tables-between-servers/
• Import / export of buffer_pool | innodb_lru_dump_restore
  • http://www.mysqlperformanceblog.com/2010/01/20/xtradb-feature-save-restore-buffer-pool/
• Transactional replication | innodb_overwrite_relay_log_info
  • http://www.mysqlperformanceblog.com/2009/03/04/making-replication-a-bit-more-reliable/
• Better handling of corrupted tables ( do not crash, but return “corrupted table” error)
• Show internal InnoDB data dictionary
  • http://www.mysqlperformanceblog.com/2010/03/20/innodb-tableindex-stats/
• Show InnoDB locking/io profiling in slow.log
• Microsecond precision for event time in slow query log | microseconds_in_slow_query_log
  • With standard microsecond slow log you can have the query time and the lock time with microsecond precision, but the event time is only written with second precision (e.g. # Time: 090402 9:23:36). This option allows you to have the event time in microsecond precision (e.g. # Time: 090402 9:23:36.123456).
• Use global scope for variables of the slow_extended patch | control_global_slow
  • The slow_extended patch adds valuable information to the slow query log and you can configure which statistics you want to collect. With this change you can force the slow logging process to use the global values for any variable of the slow_extended patch so you are sure to change the behavior of all threads without a restart. For more information on the concept, see use_global_long_query_time

Blog post on XtraDB improvements: http://www.mysqlperformanceblog.com/2009/08/13/xtradb-the-top-10-enhancements/