Using

EDB Postgres Distributed (PGD) provides multi-master replication and data distribution with advanced conflict management, data-loss protection, and throughput up to 5X faster than native logical replication . It enables distributed Postgres clusters with high availability up to five 9s.

By default, EDB Postgres Distributed uses asynchronous replication, applying changes on the peer nodes only after the local commit. You can configure additional levels of synchronicity between different nodes, groups of nodes, or all nodes by configuring Group commit , CAMO or commit-at-most-once <CAMO or commit-at-most-once> , or Eager Replication.

• Application use
  • Application behavior
    • Replication behavior
    • Truncate
    • Row-level locks
    • Sequences
    • Binary objects
    • Rules
    • Base tables only
    • Partitioned tables
    • Triggers
    • Toast
    • Other restrictions
  • DML and DDL replication and nonreplication
    • DML replication
    • DDL replication
    • Nonreplicated statements
  • Nodes with differences
    • Replicating between nodes with differences
    • Comparison between nodes with differences
    • Replicating between different release levels
  • General rules for applications
    • Background
    • Rules for applications
  • Timing considerations and synchronous replication
  • Use of table access methods (TAMs) in PGD
  • Feature compatibility
    • Server feature/commit scope interoperability
    • Commit scope/commit scope interoperability
• Node management
  • Node types
    • Data nodes
    • Witness nodes
    • Logical standby nodes
    • Subscriber-only nodes
  • Groups and subgroups
    • Groups
    • Sub-groups
  • Creating and joining PGD groups
    • Creating and joining PGD groups
  • Witness nodes
    • Witness Nodes within PGD groups or regions
    • Witness Node outside regions
  • Logical standby nodes
  • Subscriber-only nodes and groups
    • Subscriber-only nodes
    • Subscriber-only groups
  • Viewing PGD topology
    • Listing PGD groups
    • Listing nodes in a PGD group
  • Removing nodes and groups
    • Removing a node from a PGD group
    • Removing a whole PGD group
  • Joining a heterogeneous cluster
  • Connection DSNs and SSL (TLS)
  • Decoding worker
    • Enabling
    • Disabling
  • Replication slots created by PGD
    • Group replication slot
    • Hashing long identifiers
  • Node restart and down node recovery
  • Maintenance commands through proxies
    • Maintenance and performance
    • Maintenance and proxies
    • Maintenance commands
    • A note on node names
    • Finding the write leader
    • Formatting bdr.run_on_nodes() output
    • Defining a function for maintenance
• Postgres configuration
  • Postgres settings
  • Time-based snapshots
    • snapshot_timestamp
  • Max prepared transactions
    • max_prepared_transactions
• DDL replication
  • DDL overview
    • Replicated DDL
    • Differences from PostgreSQL
    • Executing DDL on PGD systems
  • DDL replication options
  • DDL locking details
    • The global DDL lock
    • The global DML lock
  • Managing DDL with PGD replication
    • Minimizing the impact of DDL
    • Handling DDL with down nodes
    • Statement-specific DDL replication concerns
  • DDL Command handling matrix
    • Command matrix
    • Command Notes
  • DDL and role manipulation statements
  • Workarounds for DDL restrictions
    • Adding a CONSTRAINT
    • Adding a column
    • Changing a column’s type
  • PGD functions that behave like DDL
• Security and roles
  • Roles
  • Role management
    • Role rule - No un-replicated roles
    • Roles for new nodes
    • Connections and roles
  • PGD predefined roles
    • bdr_superuser
    • bdr_read_all_stats
    • bdr_monitor
    • bdr_application
    • bdr_read_all_conflicts
  • Roles and replication
    • DDL and DML replication and users
    • Differing Table Ownership
    • Replication and row-level security
    • bdr_superuser role and replication
    • Privilege restrictions
    • Foreign key privileges
  • Access Control
    • Catalog tables
    • PGD functions and operators
    • Granting privileges on catalog objects
    • Triggers
• Sequences
  • PGD global sequences
    • SnowflakeId sequences
    • Timeshard sequences
    • Globally allocated range sequences
    • Converting a local sequence to a galloc sequence
  • UUIDs, KSUUIDs, and other approaches
    • UUIDs
    • KSUUIDs
    • Step and offset sequences
    • Composite keys
  • See also
• Durability
  • Introducing
  • Commit scope kinds
  • Working with commit scopes
    • An Overview of Durability options
    • Durability Terminology
    • Commit scopes
    • Commit scope rules
    • Comparing Durability Options
    • Group Commit
    • Commit At Most Once
    • Lag Control
    • PGD Synchronous Commit
    • Administering
    • Legacy synchronous replication using PGD
    • Limitations
    • Internal Timing of Operations
• Consistency
  • Conflicts
    • How conflicts happen
    • Types of conflict
    • Avoiding or tolerating conflicts
    • Conflict detection
    • Conflict resolution
    • Conflict logging
    • Data verification with LiveCompare
  • Column-level conflict detection
    • Enabling and disabling column-level conflict resolution
    • DDL locking
    • Current versus commit timestamp
    • Inspecting column timestamps
    • Handling column conflicts using CRDT data types
    • Notes
  • Conflict-free replicated data types
    • State-based and operation-based CRDTs
    • Disk-space requirements
    • CRDT types versus conflicts handling
    • CRDT types versus conflict reporting
    • Resetting CRDT values
    • Implemented CRDT data types
  • Eager conflict resolution
    • Usage
    • Error handling
    • Effects of Eager Replication in general
    • Effects of MAJORITY and ALL node replication in general
• Parallel Apply
  • What is Parallel Apply?
    • Configuring Parallel Apply
    • When to use Parallel Apply
    • Monitoring Parallel Apply
    • Disabling Parallel Apply
    • Deadlock mitigation
    • Parallel Apply Support
• Replication sets
  • Using replication sets
  • Behavior of partitioned tables
  • Behavior with foreign keys
  • Replication set membership
  • Listing replication sets
  • DDL replication filtering
  • Selective Replication Example
    • Cluster Configuration
    • Application Requirements
    • Creating tables
    • Viewing groups and replication sets
    • Adding tables to replication sets
    • Testing Selective Replication
• Application connection management
  • Configuration
    • EDB Postgres Distributed Proxy
    • Installing PGD Proxy
    • Proxies, Raft, and Raft subgroups
• Backup and recovery
  • Backup
    • pg_dump
    • Physical backup
    • Eventual consistency
    • Point-in-time recovery (PITR)
  • Restore
    • EDB Postgres Distributed cluster failure or seeding a new cluster from a backup
    • Cleanup of PGD metadata
    • Cleanup of replication origins
    • Cleanup of replication slots
• Monitoring
  • OpenTelemetry integration
    • OTEL and OLTP compatibility
    • Metrics collection
    • Tracing
    • TLS support
  • Monitoring through SQL
    • Monitoring overview
    • Monitoring node join and removal
    • Monitoring Replication Peers
    • Monitoring PGD replication workers
    • Monitoring PGD writers
    • Monitoring global locks
    • Monitoring conflicts
    • Apply statistics
    • Standard PostgreSQL statistics views
    • Monitoring PGD versions
    • Monitoring Raft consensus
    • Monitoring replication slots
    • Monitoring transaction COMMITs
• EDB Postgres Distributed Command Line Interface
  • Installing PGD CLI
    • Installing automatically with Trusted Postgres Architect (TPA)
    • Installing manually on Linux
  • Using PGD CLI
    • What is the PGD CLI?
    • Passing a database connection string
    • Specifying a configuration file
    • Specifying the output format
    • Accessing the command line help
  • Configuring PGD CLI
    • PGD CLI and database connection strings
    • Configuring the database to connect to
  • Discovering connection strings
    • PGD CLI and database connection strings
    • Getting your database connection string
  • Command reference
    • Synopsis
    • Options
    • See also
• Transaction streaming
  • PGD enhancements
    • Caveats
  • Configuration
    • Node configuration using bdr.default_streaming_mode
    • Group configuration using bdr.alter_node_group_config()
    • Configuration setting effects
  • Monitoring
• Testing and tuning PGD clusters
  • Trusted Postgres Architect
  • pgd_bench
  • Notes on pgd_bench usage
    • Performance testing and tuning
• Stream triggers
  • Trigger execution during Apply
    • Missing-column conflict resolution
    • Data loss and divergence risk
  • Terminology of row-types
  • Conflict triggers
  • Transform triggers
  • Row contents
  • Execution order
  • Stream triggers examples
• AutoPartition
  • Auto creation of partitions
  • Autopartition Examples
  • RANGE-partitioned tables
  • Stopping automatic creation of partitions
  • Waiting for partition creation
  • Finding a partition
  • Enabling or disabling AutoPartitioning
• Explicit two-phase commit (2PC)
  • Use
• Timestamp-based snapshots