This major release boasts a long awaited feature: the implementation of parallel copy for backup/recovery through the parallel_jobs option. This highly anticipated feature allows Barman to perform copies at an accelerated rate, speeding up the backup and recovery process for the rsync copy method.
Subito, long time partners of 2ndQuadrant and sponsor of the Barman project, participated in early testing of the parallel copy feature.
This feature has been a goal of the development team for some time. "Parallel copy fills an important gap for Barman in very large database (VLDB) scenarios, by giving users flexibility and freedom in architecture design to help them reduce both backup and recovery times sensibly," said Gabriele Bartolini, one of the lead developers behind Barman. "This feature takes advantage of the code refactoring activity that has been previously performed with 2.0 and 2.1 releases, including the thousands of automated tests that help us make Barman more robust."
Other notable features in this major release include:
- Support custom WAL size (PostgreSQL 8.4 and up)
- Improve check command
- Add external_configuration to backup_options
- Add --get-wal and --no-get-wal options to barman recover
- Add max_incoming_wals_queue global/server option for the check command
- Documentation improvements
Additionally, several minor bug fixes have also been implemented. Read the full release notes for Barman 2.2
here.
Navionics, a manufacturer of electronic navigational charts and founding sponsor of the Barman project, migrated its systems to PostgreSQL in 2012. With several servers under Barman backup, the Navionics team stated that they are proud to have contributed again in the developing of a new Barman release that includes parallel jobs.
"We have been a sponsor of Barman since the first release because we need a reliable system and disaster recovery [capability] for our production servers," stated the Navionics DBA Team.
"We are sure that [parallel copy within Barman] will allow us to have faster backup and restore for our huge databases."