One of the most frustrating challenges we hear from developers is the burden of traditional infrastructure management. Provisioning infrastructure, scaling alongside deployments, and managing upgrades and versioning is complicated, time consuming, and wastes resources for teams. With this setup, you always pay for compute whether or not there's work to do.
Introducing Serverless Workers for Cloud Run#
That's why today, we're announcing Temporal Serverless Workers support for Google Cloud Run, which makes it simpler to run and scale your workers. Temporal Serverless Workers removes the need to perform infrastructure planning and develop an autoscaling strategy when getting started with Temporal.
Temporal Serverless Workers are the fastest way to launch your first workflow on Temporal; to learn more, see here: getting started.
Why Cloud Run Worker Pools are a natural fit#
Google Cloud Run is a fully managed application platform for running your code, function, or container on top of Google's highly scalable infrastructure. Temporal has long supported running Workers on Cloud Run, and this effort builds on the expertise we've developed through our partnership with GCP.
Google's recent addition of Cloud Run Worker Pools, which are designed for long-running tasks or background applications, provide a great fit for the operational model of Temporal Workers. Serverless Workers integrates with Cloud Run Worker Pools and provides a custom, highly responsive, and efficient autoscaling algorithm that allows you to skip the manual setup and tuning that was previously required.
How the autoscaling works#
Temporal watches key metrics on the server that indicate work is queueing up, and dynamically adjusts capacity within your Worker Pools, without the need to set up your own metrics pipeline, configure your own autoscaling triggers, or deal with the lag inherent to metrics-based autoscaling approaches.
Writing your worker code and activities is exactly the same when using Serverless Workers, and you’ll deploy your worker to Cloud Run using the standard pathway of creating a Docker container, configuring your Worker Pool specifications, and creating the Worker Pool.
Once the Worker Pool is deployed, tell Temporal about it by creating a Worker Deployment and Worker Deployment Version, passing in the resource name and IAM role. From here, Temporal takes control of all scaling activities, even scaling to zero when appropriate.
After starting workflows, we can see that the Worker Pool is running workers to process the tasks in the workflow:
In response to starting our workflows, work is added to the backlog on Temporal Cloud, and our Worker Pool is scaling up to accommodate the demand in the Cloud Run Worker Pool metrics:
Get started#
Serverless Workers for Google Cloud Run is available as a pre-release product as of today. To get access, reach out to your account manager or use the resources below to learn more.