Microsoft SQL Server on Red Hat Enterprise Linux offers instance- and database-level high availability (HA) through integration with Pacemaker, a high availability cluster resource manager optimized
for Red Hat Enterprise Linux. With orchestration capabilities for monitoring, failure detection, and automatic failover, Pacemaker detects and recovers applications and VMs in the event of node- or
resource-level failures. Red Hat provides detailed Pacemaker configuration information to optimize operations, while SQL Server uses unique Pacemaker-specific resource agents to implement instance-
and database-level HA. Microsoft also maintains a separate package repository for open source,
Pacemaker-specific agents for SQL Server.
Additionally, SQL Server supports instance-level HA and redundancy with Always On Failover Cluster
Instances (FCI) to ensure SQL Server instances remain available during planned and unplanned
outages. An FCI comprises two or more cluster nodes with access to storage area network (SAN)
or direct-attached cluster shared storage. Only one node is active at any time. Secondary nodes
are passive but ready to assume the role of active node during failover. Pacemaker manages all FCI
cluster resources running on Red Hat Enterprise Linux.
SQL Server achieves database-level HA and disaster recovery with Always On Availability Groups.
These groups support a replicated environment for a discrete set of user databases, known as availability databases, running on similarly configured hardware. Several types of non-exclusive availability
groups are possible with SQL Server and Red Hat Enterprise Linux:
• Availability groups configured for high availability and disaster recovery are sets of databases that fail over together. They provide high availability, disaster recovery, and read-scale balancing using Pacemaker as the cluster manager.
• Availability groups configured for read-only, scale-out workloads are sets of databases that are replicated to other instances of SQL Server for applicable workloads. They do not require a
• Distributed availability groups span two separate availability groups for failover and disaster recovery purposes. Underlying availability groups can be configured on the same platform and managed
by Pacemaker — or placed on different platforms for migration and disaster recovery purposes.
• Cross-platform availability groups include one or more replicas on both Microsoft Windows and
Red Hat Enterprise Linux, providing multiplatform support and minimizing application disruption for
database migrations between Windows and Linux.
Each availability group can have one primary replica and up to eight secondary replicas. Other capabilities include multidatabase failover, multiple synchronous and asynchronous secondaries, manual or
automatic failover, and active secondaries for read and backup workloads.