Primary/backup replication (also known as primary copy replication master-slave replication or log shipping) is perhaps the most commonly used replication method, and the most basic algorithm.
- asynchronous primary/backup replication and
- synchronous primary/backup replication
P/B is very common. For example, by default MySQL replication uses the asynchronous variant. MongoDB also uses P/B (with some additional procedures for failover). All operations are performed on one master server, which serializes them to a local log, which is then replicated asynchronously to the backup servers.
What is key in the log-shipping / primary/backup based schemes is that they can only offer a best-effort guarantee (e.g. they are susceptible to lost updates or incorrect updates if nodes fail at inopportune times). Furthermore, P/B schemes are susceptible to split-brain, where the failover to a backup kicks in due to a temporary network issue and causes both the primary and backup to be active at the same time.
To prevent inopportune failures from causing consistency guarantees to be violated; we need to add another round of messaging, which gets us the two phase commit protocol (2PC).
- Replication techniques for availability - Robbert van Renesse & Rachid Guerraoui, 2010
- Concurrency Control and Recovery in Database Systems