Geographically distributed PostgreSQL cluster with very high availability and conflict resolution
For businesses that require geographically distributed multiple masters, 2ndQuadrant’s solution for AlwaysOn Multi-Master Cluster Evolution is based on years of experience fulfilling enterprise needs. The architecture relies on best practices, provides high levels of reliability for production use, and is based on ground-breaking Postgres-BDR technology – exclusively developed by 2ndQuadrant.
Geographically Distributed Clusters
The Solution is specifically designed for geographically distributed databases, providing the flexibility to choose between “eager” and “efficient” modes. The following characteristics can be tweaked based on your latency-consistency needs:
• Tolerance of network latency
• Latency for read/write access to data
• Replication lag
This solution presents true high availability for your PostgreSQL database(s) providing up to 5 nines of availability. It guarantees an AlwaysOn service, regardless of node failure, with these unique features:
Shadow Masters. A second master working in passive, receive-only mode that is ready to accept write connections at any point. This allows the cluster to switchover without the application noticing any delays.
Automatic Failover. In case of failure or loss of a master node, The Solution allows a quick promotion of a standby to a master, preventing any service interruption.
Tolerance of Software Failure. Logical and physical replication can run simultaneously, so the cluster is protected by two kinds of replication; providing higher tolerance to software failure.
The Cluster Evolution Solution gives you the advantage of keeping your PostgreSQL instances updated all the time. There is no need of matching configuration, CPU architecture, or software versions among the master nodes. This allows the solution to perform one-by-one upgrades with zero downtime when a new PostgreSQL release or update is available.
Point-in-Time Recovery (PITR)
This Solution allows any node to be backed up from any node in a cluster and then restored using Point-In-Time Recovery. Each node can perform backups, ensuring that whole-cluster backups are taken in every region to guarantee full disaster recovery capability.