Lesson 1 of 5
High availability versus disaster recovery
9 min
Listen to this lesson
High availability (HA) and disaster recovery (DR) are often mentioned in one breath, but they protect against different things. HA keeps a system running through the failure of a single component, such as a server, a process or a disk, usually within one site and with little or no data loss. DR restores service after the loss of a whole site or region, usually at a distant location, and accepts some data loss and a longer outage. A landscape needs both, and neither replaces backups, which protect against the failures that replication faithfully copies, such as a wrong mass update or corrupted data.
Single points of failure in an ABAP system
| Component | Why it matters | Typical HA measure |
|---|---|---|
| ABAP central services (ASCS): message server and enqueue server | One per system; without it, logons and locks stop | Cluster failover of ASCS, with the enqueue replication server (ERS) keeping a copy of the lock table |
| Application server instances | Users and jobs run here | Several instances (N+1) behind logon groups, so losing one is tolerable |
| SAP HANA database | All data lives here | SAP HANA system replication to a standby, often with cluster-controlled takeover |
| Shared file systems and network | Transport directory, profiles, interfaces | Redundant storage and network paths |
Where an ABAP system can fail and the usual HA answer. Architecture details vary by platform.
HA and DR both copy changes to another place, which is why they cannot fix logical errors. If Ana accidentally posts 2,000 duplicate supplier invoices, replication sends every one of them to the standby within seconds. Only a point-in-time restore from backups and logs, or a business correction, helps then. Keep that in mind when someone says the system is safe because it is replicated.
Testers meet these layers more often than they think. A test system without HA restarts after a host failure and loses the work in flight; a quality system that is a restored copy of production shows data as of the backup time; and a test of a long-running job may be interrupted by a planned takeover in a maintenance window. Knowing which layer is in play explains results that otherwise look like application defects.
Who does what under RISE
Contoso's CMP runs in SAP S/4HANA Cloud Private Edition under RISE. In SAP's roles and responsibilities document (v.3-2025; later versions such as v7-2025 exist, so check the one attached to your contract), managing standby databases with SAP HANA system replication for high availability is a standard service for productive systems, but only if dedicated standby databases are in the contract, and the document notes that failover tests are not performed regularly per system. Disaster recovery is an optional service, provided only if it is in the contract. Backups and their monitoring are standard services. Developing DR procedures for the parts in the customer's responsibility, such as interfaces, connectivity and organizational measures, is an excluded task: Contoso's own job.
| Area | SAP ECS (per R&R, if contracted) | Contoso |
|---|---|---|
| Backups | Database and log backups per policy; monitoring of backups | Decide retention needs; validate restored data |
| HA | Manage HANA system replication standby for production, if in contract | Design expectations; test what users experience |
| DR | Optional service: set-up, replication monitoring, failover execution in a DR test | DR procedures for interfaces and organization; functional tests |
| Sign-off | Executes technical steps | Signs off that recovery works for the business |
What would you do?
In a steering meeting, someone says Contoso does not need to discuss disaster recovery because 'RISE includes high availability'. Grace asks Sam to respond. What is the most accurate answer?
- HA protects against component failures nearby with near-zero data loss; DR protects against losing a site.
- Backups protect against logical errors that HA and DR replicate faithfully.
- Single points of failure: ASCS (with ERS), application servers (N+1), SAP HANA (system replication), storage and network.
- Under RISE, the HA standby is a standard service if contracted, DR is optional, and customer-side DR procedures stay with the customer.
Check your understanding · not graded
Lesson check
1. Which failure does disaster recovery address that high availability does not?
2. Why can system replication not protect against a mistaken mass update?
3. What protects the lock table when the ASCS instance fails over?
4. Under SAP's RISE R&R, how is disaster recovery classified?