Can the platform support automated failover, recovery, and workload restart?

Summary

• Yes. The Databricks Platform supports automated failover, recovery, and workload restart across multiple layers, from availability-zone failures inside a region to full cross-region disaster recovery.
• For an availability-zone failure, the control plane fails over automatically to healthy zones, typically within a 15-minute RTO and with an RPO of 0 (no committed data lost).
• Serverless compute provides automatic multi-zone failover with no extra configuration; classic compute can restart a job and cluster in another availability zone and request replacement nodes when a zone or driver is lost.
• Jobs can be configured with automatic retries and exponential backoff to recover from transient failures without manual intervention.
• For a full region outage, Databricks Managed Disaster Recovery continuously replicates data and configuration to a secondary region and lets you trigger a failover that swaps to a stable URL and resumes job schedules in the new region.

Can the platform support automated failover, recovery, and workload restart?

Yes. The Databricks Platform provides automated resilience at several layers. High availability (HA) handles a failure confined to a single availability zone automatically, while disaster recovery (DR) handles the loss of an entire region through a controlled, automatable failover to a secondary region. Job-level retries recover individual workloads from transient failures. Together these keep committed data safe and get work running again within your recovery targets.

Why the Databricks Platform recovers automatically

• Automatic control-plane failover. In a region with multiple availability zones, the control plane is distributed across zones and fails over automatically to healthy zones, typically within a 15-minute RTO and with an RPO of 0, so no committed data is lost. During a zonal outage, traffic is redirected from the failed zone with no impact to external-facing IPs or hostnames. See Design a high availability strategy.
• Serverless compute fails over on its own. serverless compute provides automatic multi-zone failover with no additional configuration; Databricks manages zone selection and replaces VMs lost to a zonal outage, redistributing workloads automatically.
• Classic compute restart in another zone. For classic compute, enabling the auto-AZ option lets the cluster manager restart a job and its cluster in a different availability zone if the original zone fails or the driver node is lost, and Databricks distributes cluster nodes across zones and requests replacement nodes when one is lost. Deploy workspace subnets across at least two availability zones (three recommended). See intra-region high availability guarantees.
• Automatic job retries. Configure job retries with exponential backoff so workloads recover automatically from transient failures, including brief zone degradation, without manual restarts.
• Cross-region disaster recovery. For a full region outage, you fail over to a secondary region. Databricks Managed Disaster Recovery continuously replicates Unity Catalog metadata, managed table data, and workspace assets to the secondary region, provides a stable workspace URL that survives failover, and lets you trigger failover as a single operation that reverses replication, swaps the stable URL, and pauses job schedules in the former region while resuming them in the new one. See how Databricks Managed Disaster Recovery helps Capital One achieve lakehouse resilience.
• Match protection to your objectives. Set a Recovery Time Objective (RTO) and Recovery Point Objective (RPO) per workload, then choose active-passive, active-active, or backup-and-restore. Databricks recommends testing recovery procedures regularly so real recovery matches your targets. See the disaster recovery guide and reliability best practices.

Getting started

• Set an RTO and RPO for each workload, then read Design a high availability and disaster recovery strategy.
• Choose serverless compute for automatic multi-zone resilience, or spread classic compute subnets across at least two availability zones and enable auto-AZ.
• Configure job retries with exponential backoff, and plan cross-region recovery with the disaster recovery guide.
• Test your recovery procedures regularly.

FAQs

Does Databricks fail over automatically if an availability zone goes down?

Yes. In a multi-zone region, the control plane fails over automatically to healthy zones, typically within a 15-minute RTO and with an RPO of 0, and serverless compute fails over across zones with no additional configuration.

Will my jobs restart automatically after a failure?

Classic compute can restart a job and cluster in another availability zone when a zone or driver node is lost, and you can configure job retries with exponential backoff so workloads recover automatically from transient failures.

How does recovery work if an entire region goes down?

You fail over to a secondary region. Databricks Managed Disaster Recovery continuously replicates metadata, managed table data, and workspace assets, and lets you trigger a failover that swaps to a stable URL and resumes job schedules in the new region.

How do I decide how much automated resilience a workload needs?

Set the workload's RTO and RPO, then choose a pattern (active-active for the strictest targets, active-passive for important workloads, and backup-and-restore for non-critical ones) and test recovery procedures regularly.