Overview

SingleStore High Availability (HA) is a robust feature designed to maintain continuous data access and system resilience in distributed database environments. It achieves this by replicating data across multiple nodes, ensuring that in the event of hardware or software failures, the system can seamlessly continue operations without significant downtime. The architecture employs availability groups, each containing primary and replica partitions. This setup allows for automatic failover, where replica partitions are promoted to primary status if their counterparts fail, ensuring data availability and consistency. Additionally, SingleStore supports multi-availability zone (AZ) deployments, providing geographic redundancy to protect against regional outages. Advanced features like Point-in-Time Recovery (PITR) and Smart Disaster Recovery (SmartDR) further enhance the system's ability to recover from failures, allowing for precise restoration of data and continuous replication across regions. These capabilities make SingleStore HA suitable for mission-critical applications requiring high uptime and data integrity.

Features and Capabilities

• Availability Groups: Data is stored redundantly across nodes in availability groups, each containing both primary and replica partitions. This structure ensures that if a primary partition fails, a replica can take over without interrupting service.
• Automatic Failover: In case of node failure, SingleStore automatically promotes replica partitions to primary status, minimizing downtime and maintaining data availability.
• Multi-AZ Protection: Enterprise plans offer multi-availability zone failover, distributing data across independent data centers to safeguard against local and regional failures, including natural disasters and power outages.
• Point-in-Time Recovery (PITR): Allows for the restoration of databases to a specific point in time, down to the microsecond, providing flexibility in data recovery and backup management.
• Smart Disaster Recovery (SmartDR): Provides continuous asynchronous replication of data between primary and geographically separate secondary regions, ensuring protection against major regional outages.
• Zero-Config Setup: SingleStore HA is enabled by default in SingleStore Helios, requiring no manual configuration, simplifying deployment and management.
• Load Distribution: Balanced distribution of replica partitions across nodes ensures that failover does not overload any single node, maintaining system performance during recovery.
• Redundancy Levels: Supports redundancy-1 and redundancy-2 configurations, allowing users to choose the level of data replication that best fits their availability requirements.
• Cluster Rebalancing: Automatic rebalancing of data partitions when nodes are added or removed ensures even distribution of data and optimal performance across the cluster.
• Disaster Recovery Commands: Provides commands to manage high availability and disaster recovery operations, including attaching and detaching leaf nodes, and rebalancing partitions to maintain cluster health.