How long do HA initial synchronization, failover, and switchover take?
Applicable Products
- QuTS hero h5.3.0 or later
- High Availability Manager
Background
High availability (HA) allows two QNAP NAS devices to form an active-passive setup for uninterrupted service. The active node provides services while the passive node continuously syncs data from the active node, ready to take over any time. This transition is called a failover if triggered by a failure, and a switchover if triggered by non-failure events.
When synchronizing data during HA cluster creation, the system needs to replicate a large amount of data from one node to the other. For failovers or switchovers, the system needs to ensure seamless continuation of applications and services. The time required for these operations can vary significantly depending on hardware specifications, storage configuration, and network conditions.
Details
The actual time required for initial synchronization and for a failover or switchover depends on the following factors:
- Data volume: Larger amounts of data increases synchronization time.
- Storage performance: HDDs are slower than SSDs. RAID configuration also affects performance.
- Network conditions: The heartbeat connection speed affects how fast data is transferred (we recommend using ports and a direct cable connection of 10 GbE or faster).
- System load: Ongoing services and background tasks may consume resources and slow down the process time.
- Applications and services: It takes time to shut down and restart each application and service, especially in an environment with iSCSI LUNs.
Typical Time Ranges (Reference Only)
| Process | Typical Time Range |
|---|---|
| Initial synchronization | Tens of minutes to several hours |
| Failover/switchover | Around 30 seconds to several minutes |
| Failover/switchover with multiple iSCSI LUNs | Several minutes to tens of minutes The more iSCSI LUNs there are, the longer it takes. |
The actual time required depends on your environment and system configuration. We recommend performing switchover tests during non-production hours to evaluate performance in your specific setup.
Further Reading
How can I reduce initial synchronization time for my HA cluster?
