Migrating traffic

Traffic migration switches the request routing between the versions within a service of your application, moving traffic from one or more versions to a single new version.

For information about splitting traffic between two or more versions of your app, see Traffic Splitting.

Before you begin

Before you can configure traffic to a version, ensure that your user account includes the required privileges.

Migrating traffic gradually

You can choose to route requests to the target version, either gradually or immediately.

By default, warmup requests are disabled and traffic is migrated immediately to a version. You can also choose to disable warmup requests if you want the traffic immediately migrated to a version.

If you immediately migrate traffic to a new version without any running instances then you will experience a spike in latency for loading requests. Deploying a new version with the same name as an existing version causes an immediate traffic migration. All instances of the old version are immediately shut down. There will be a latency spike due to loading requests for the new version.

If warmup requests are enabled, you can migrate traffic between versions that reside in different environments only by specifying to migrate traffic immediately.

Adding warmup requests to your application

When warmup requests are enabled, traffic is migrated gradually by first sending a warmup request to new instances before those instances receive any user requests. Warmup requests improve user response time by allowing the version currently receiving traffic to handle those requests but the traffic migration to the new version can take a short amount of time while the new instances are created.

When warmup requests are not enabled, user requests are sent to those new instances before they have been created. Due to the delay caused by creating the new instances and loading application code, latency can occur for those user responses.

For all runtimes except for Java, to avoid latency and enable warmup requests, include the inbound_services element in your configuration file before you deploy your application to App Engine.

For example, you include the following in your app.yaml file before deploying it to App Engine:

- warmup

For complete details about enabling warmup requests, see Configuring warmup requests to improve performance.

Migrating traffic to a new version


To migrate traffic in the Google Cloud console, go to the Versions page:

Go to Versions

  1. Select the version to which you want to migrate 100% of the traffic.
  2. Click Migrate traffic.
  3. Optional: When warmup requests are enabled your traffic is migrated gradually. To migrate traffic immediately, select the option under the Show advanced options section.


After installing the Google Cloud CLI, you run the gcloud app services set-traffic command to migrate 100% of traffic to a single version. For example:

  • To migrate traffic immediately:
    gcloud app services set-traffic [MY_SERVICE] --splits [MY_VERSION]=1
  • To gradually migrate traffic, you include the optional --migrate flag:
    gcloud app services set-traffic [MY_SERVICE] --splits [MY_VERSION]=1 --migrate


To programmatically migrate traffic, you can use the Admin API, see Migrating and Splitting Traffic for details.