What Is Kelda: FAQ

Kelda is a development environment for Kubernetes. It’s a drop in replacement for Docker Compose that allows you to quickly test your code in a remote environment that matches production, without the complexity of interacting with Kubernetes directly.

Architecture diagram

How does it work?

Architecture diagram

The Kelda CLI (which runs on your laptop) handles code syncing and setting up tunnels.

The CLI talks to an operator running in your Kubernetes cluster, which creates a development namespace for you containing all your services.

Check out our architecture documentation for more information.

What’s the development workflow?

  1. Run kelda dev <service> to start working on a service.

Kelda will start code syncing for the service, and boot the production versions of all the other services.

  1. Edit your code locally.

Kelda works with any editor – it watches for file changes at the filesystem level.

Once Kelda notices a file change, Kelda automatically syncs the new file over, and restarts the remote process.

  1. Test your code using tunnels.

Tunnels connect the ports on your local machine to the services in your development environment. These tunnels can be used with everything from your browser, to Postman, to your database client.

  1. If necessary, debug.

Kelda makes it easy to find all the information you need to quickly debug your code.

Get logskelda logs <service>
Get a shell in a containerkelda ssh <service>
Debug Kubernetes issueskubectl directly

For more advanced debugging, you can hook up your language’s debugger, or use Telepresence to run your service locally.

To learn more, try Kelda out yourself with an example application on our demo cluster, or watch a video of Kelda in action.

What configuration does it need?

To use Kelda, you have to tell it how to boot your services. Kelda just needs the Kubernetes YAML that you would use in production.

Kelda also requires a configuration file per service to enable file syncing when you’re developing on that service.

Check out our configuration guide to learn more.

Why Kelda?

You might like Kelda if…

  • Your laptop is slow because it doesn’t have enough CPU and RAM to run all your services.
  • You’re using Kubernetes in production and would like your development environment to run on Kubernetes as well.
  • You don’t want to deal with kubectl during development. Testing your code takes too long because you have to wait for Docker to do a full image build and push.

How is Kelda different than Minikube?

Minikube, k3s, and Docker for Mac all provide a local instance of Kubernetes, but don’t directly address the development workflow.

If you’re using one of these tools for local development, you have to manually boot your services, and interact directly with them via kubectl. Plus, deploying changes by building and pushing images is very slow.

Kelda is compatible with all of these Kubernetes offerings if you’d like to run your Kubernetes cluster locally, but automate the workflow with Kelda.

How is Kelda different than Docker Compose?

Kelda runs your services remotely, making your laptop much faster. In contrast, Docker Compose runs everything on your local machine.

Plus, with Kelda, both your development environment and production environment run on Kubernetes, leaving little room for error and bugs to enter your staging environment. While Docker Compose requires maintaining two completely different sets of configurations.

How is Kelda different than Skaffold?

Skaffold is for developers that prefer to work directly with Kubernetes (via kubectl). While kubectl has a lot of power, most of it isn’t necessary during development, and it requires deep Kubernetes knowledge to leverage it.

For example, Kelda has a UI that shows the status of all your microservices, while with Skaffold you have to piece together what’s going on with commands like kubectl get pods and kubectl get deployments.

How is Kelda different than Telepresence?

Kelda can work in conjunction with Telepresence. Telepresence solves the specific problem of proxying local code to a remote cluster, while Kelda also boots your dependencies, and provides a simple interface to Kubernetes.

We have an example of a Kelda Telepresence integration.

How does code syncing work for compiled languages?

We recommend compiling locally, then syncing the resulting binary into the container.

What security permissions does it need?

Kelda doesn’t open any public ports. The only ports it opens is on localhost for tunnels.

The Kelda CLI communicates directly with the Kubernetes API server to create tunnels, communicate with the Kelda Operator, and get logs.

Authentication with the API server is handled by the standard Kubernetes methods.

How much does Kelda cost?

Kelda is free for individual developers, and everything you need to get started is in the docs. If you’re a team looking to adopt Kelda, get in touch with us to discuss plans.

How does Kelda fit into my CI/CD pipeline and other environments?

Kelda focuses on the development workflow that comes before you make a PR. Once you commit your code, your existing CI/CD pipeline should take over. Using Kelda doesn’t require you to change your CI/CD pipeline or production infrastructure at all.


How do I create a Kubernetes cluster to run Kelda?

We recommend using Hosted Kelda so that you don’t have to deal with creating and maintaining a Kubernetes cluster.

Kelda also works alongside non-Kelda services, so if you already have a staging cluster, you can just deploy Kelda there.

For a list of all the options see our Kubernetes deployment guide.

How does Kelda integrate with my IDE?

Kelda works at the file system level, so it will sync code changes regardless of what IDE you used to make them.

For debuggers, just create a Kelda tunnel to the debugger port, and point your IDE at the tunnel.

Does Kelda work with templating tools like Kustomize and Helm?

Yes, Kelda supports deploying the output of a script. So you can install Kustomize locally, and reference the kustomize build command in your workspace.yaml.

We also have first class support for some tools, like Helm.

Couldn’t find what you’re looking for?

Try our docs search, or join the Kelda Slack.

Ready to try it?

Try out our example application on our demo cluster to experience Kelda yourself.

Then, once you’re ready to setup Kelda for your application, check out our Quickstart.