[Serverless Knative] Knative Docs - Scaling to Zero

Posted on 2021-12-11 Edited on 2024-09-12 In Cloud Native , Cloud Computing , Serverless , Knative Views: Word count in article: 2.2k Reading time ≈ 2 mins.

Scaling to Zero

Remember those super powers 🚀 we talked about? One of Knative Serving’s powers is built-in automatic scaling (autoscaling). This means your Knative Service only spins up your application to perform its job – in this case, saying “Hello world!” – if it is needed; otherwise, it will “scale to zero” by spinning down and waiting for a new request to come in.

What about scaling up to meet increased demand?

Knative Autoscaling also allows you to easily configure your service to scale up (horizontal autoscaling) to meet increased demand as well as control the number of instances that spin up using concurrency limits and other options - https://knative.dev/docs/serving/autoscaling/concurrency/, but that’s beyond the scope of this tutorial.

Let’s see this in action! We’re going to peek under the hood at the Pod in Kubernetes where our Knative Service is running to watch our “Hello world!” Service scale up and down.

Run your Knative Service

Let’s run our “Hello world!” Service just one more time. This time, try the Knative Service URL in your browser http://hello.default.127.0.0.1.nip.io, or you can use your terminal with curl.

1	$ curl http://hello.default.127.0.0.1.nip.io

You can watch the pods and see how they scale to zero after traffic stops going to the URL.

1	kubectl get pod -l serving.knative.dev/service=hello -w

Warning

It may take up to 2 minutes for your Pods to scale down. Pinging your service again will reset this timer.

Expected output:

NAME                                     READY   STATUS
hello-world                              2/2     Running
hello-world                              2/2     Terminating
hello-world                              1/2     Terminating
hello-world                              0/2     Terminating

Scale up your Knative Service

Rerun the Knative Service in your browser http://hello.default.127.0.0.1.nip.io, and you will see a new pod running again.

Expected output:

NAME                                     READY   STATUS
hello-world                              0/2     Pending
hello-world                              0/2     ContainerCreating
hello-world                              1/2     Running
hello-world                              2/2     Running

Exit the watch command with Ctrl+c.

Some people call this Serverless 🎉 🌮 🔥 Up next, traffic splitting!