Google Cloud Platform has launched Cloud Run, a serverless compute platform that has built-in Kubernetes containers.
The move adds to the Google Cloud serverless compute stack and leverages the company's strength in containerized platforms and notably Kubernetes. Google also has integrated Cloud Run with the rest of its portfolio, so it comes in two flavors: As a managed service and a developer environment.
Oren Teich, director of product management at Google, said in an interview that the aim for Cloud Run was to make it easier to run more workloads. "Serverless means a million things to a million people. For some it's a great dev experience. For others it's compute and a pricing model. We wanted to focus on how to make serverless work better with a platform," said Teich. "Customers want to do more things and real use cases."
Also: Top cloud providers 2019: AWS, Microsoft Azure, Google Cloud; IBM makes hybrid move; Salesforce dominates SaaS | Google Cloud Platform: An insider's guide (TechRepublic download) | Vendor comparison: Microsoft Azure, Amazon AWS, and Google Cloud (Tech Pro Research)
When it comes to serverless compute, Google Cloud has staked out an approach that's more akin to functions as a service. Serverless compute, initially defined by AWS as turnkey infrastructure that bundles management, has grown in popularity, and the appeal is that companies can drop time spent on provisioning servers to focus more on functions. Google has staked out containers and functions to go along with serverless compute. Microsoft Azure has a similar approach.
Cloud Run will operate on Google Kubernetes Engine and has built-in Knative, an API and runtime environment. For good measure, Google said it is building out second-generation runtimes for Cloud Functions and App Engine, open-sourced a Functions Framework, and connected its serverless stack to private Google Cloud resources.
Teich said that the portability will enable Cloud Run to handle any stateless Docker container in a seamless way. The key word for enterprises is stateless. An ERP system wouldn't work because it writes to disk. A legacy Java server, document processing software, and statistical analysis tool would play into Cloud Run easily and scale, he added.
- Amazon AWS: Complete business guide to the world's largest provider of cloud services
- How 'cloud-native' applications are transforming IT, and why it took so long
- Microsoft Azure: Everything you need to know about Redmond's cloud service
- The rise of Kubernetes epitomizes the transition from big data to flexible data
- Data centers: The future is software-defined
Google said Cloud Run, which includes provisioning, configuring, scaling, and management, is targeting developers that want the ease of serverless with the portability from containers. Cloud Run will be supported by Datadog, NodeSource, GitLab, and StackBlitz.
The use of Knative is designed to ensure developers can start with Cloud Run and move to Google Kubernetes Engine. Updates to Cloud Functions includes:
- New language runtimes for Node.js 8, Python 3.7, and Go 1.11 and others
- Functions Framework available in Node.js.10
- Serverless VPC Access
- Per-function identity
- Scaling controls in beta
App Engine will also be updated with new languages and Serverless VPC access.