Application Switchover Overview

Application Switchover is a migration mode for moving producer and consumer applications from a source cluster to an already-prepared destination cluster. Unlike a replication migration, switchover does not copy topic data between clusters. It provisions the destination (topics, schemas, ACLs, and consumer groups) and then coordinates a controlled, client-side cutover of your applications.

Caution

A switchover copies no data. Records that are not consumed from the source before consumers move to the destination cannot be read on the destination — they are abandoned. The cutover model and the drain gate exist to prevent this; read Choosing a Cutover Model before you begin.

When to Use Switchover

Use Application Switchover when the destination cluster will be populated by your producers going forward and you do not need historical source data replicated — for example, when standing up a new Confluent Cloud cluster for live traffic and accepting that in-flight source data is drained by existing consumers rather than copied.

If you need historical data moved to the destination, use a replication migration job instead.

Lifecycle

A switchover job moves through a dedicated set of statuses, distinct from the replication job statuses. The action available on the job’s Overview tab depends on the current status.

Status badge	Meaning	Available action
Configured	Job created; destination not yet provisioned	Provision destination
Provisioning	Destination provisioning is running (asynchronous)	— (in progress)
Provisioned	Destination is fully provisioned and ready	Begin cutover, Sync now
Cutover in progress	Cutover has started; clients are being moved	Mark complete (in the cockpit)
Completed	Switchover completed; monitoring stopped	— (record only)
Failed	Provisioning failed	Retry provision, Sync now

Status flow: Configured -> Provisioning -> Provisioned -> Cutover in progress -> Completed

Provisioning failures move the job to Failed, from which it can be retried after fixing the configuration.

Tip

The displayed badge differs from the internal sub-status: a newly created job’s internal status is Pending but the badge reads Configured. On completion the badge reads Completed, while the derived top-level JobStatus becomes Stopped underneath (used for dashboard sorting and filtering) — two different fields, so the badge you see says Completed.

Tip

Editing the job’s selections or clusters is locked once cutover begins (CutoverInProgress) and after it Completes. A Failed job stays editable so you can fix the configuration and reprovision.

The Switchover Journey

1. Create the job — choose Application Switchover in the wizard and select the provision targets (topics, consumer groups, ACLs). You can also start from the migration planner.
2. Provision the destination — create the destination topics, ACLs, and consumer-group offsets, and optionally schemas.
3. Choose a cutover model — pick the granularity and order that match your downtime and data-loss tolerance.
4. Begin cutover — scan the source for producers and consumer groups, then acknowledge and start.
5. Work the cutover cockpit — follow the guided steps, watch the live flow, and respect the drain gate.
6. Complete the switchover — once every consumer group has cleared the source and producer cutover is acknowledged, mark it complete.