High Availability and Disaster Recovery
Availability within a region
HA design uses availability zones for the database tier, since zonal deployment protects against datacenter failure, and availability sets where zones are unavailable. For SAP central services, use a Windows or Linux cluster with a file share witness or Azure shared disk, and for HANA use HANA system replication with automatic failover managed by a cluster agent. Pacemaker on SUSE or RHEL with the Azure fencing agent is the standard pattern; STONITH must be configured so a failed node cannot corrupt shared state.
Disaster recovery across regions
DR typically pairs a primary region with a secondary region using HANA system replication in async mode, plus Azure Site Recovery for application servers. Traps include assuming that zone redundancy equals cross-region DR, and forgetting that the secondary region needs its own networking, storage, and quota reservations. Another trap is failover direction and replication mode: sync replication is intra-region, async is inter-region. Test recovery plans and document RTO/RPO; the exam often asks which component satisfies a stated RPO.
Worked example
A requirement states RPO of five minutes and RTO of one hour across regions. Rule of thumb: use async HANA system replication to a secondary region (async can meet minutes-level RPO), pre-provision the secondary infrastructure, and automate failover with a tested runbook. Zone redundancy alone does not satisfy a cross-region requirement.