Serverless

Deploying applications as serverless services is a good alternative to a regular Deployment / Service model when the applications purpose is to serve HTTP or gRPC requests, either internal or from the Internet. CoreWeave uses the KNative runtime to support deploying a serverless application with a single manifest. No installation necessary.
Serverless Benefits
Automatic public HTTPS endpoints
Auto-scaling, including scale to zero
No public IP charges
​Canary deployments and other advanced deployment strategies
One Step Example
helloworld-ksvc.yaml
1
apiVersion: serving.knative.dev/v1 # Current version of Knative
2
kind: Service
3
metadata:
4
  name: helloworld # The name of the app
5
spec:
6
  template:
7
    metadata:
8
      annotations:
9
        autoscaling.knative.dev/minScale: "0" # Allow scale to Zero
10
        autoscaling.knative.dev/maxScale: "10" # Maximum number of Pods allowed to auto-scale to
11
    spec:
12
      # The container concurrency defines how many active requests are sent to a single
13
      # backend pod at a time. This configuration is important as it effects how well requests
14
      # are load balanced over Pods. For a standard, non-blocking web applocation this can usually
15
      # be rather high, ie 100. For GPU Inference however, this should usually be set to 1.
16
      # GPU Inference only processes one request at a time, and one wants to avoid a queue being
17
      # built up in the local pod instead of centrally in the Load Balancer.
18
      containerConcurrency: 10 
19
      containers:
20
        - image: gcr.io/knative-samples/helloworld-go # The URL to the image of the app
21
          resources:
22
            limits:
23
              cpu: 2
24
              memory: 4Gi
25
          env:
26
            - name: TARGET # The environment variable printed out by the sample app
27
              value: "Go Sample v1"
28
          
Copied!
After applying the manifest, get the public URL of the service. The service will be scaled up and down based on demand, and scaled to zero consuming no resources and incurring no billable charges when idle.
1
$ kubectl get ksvc
2
NAME            URL                                                       LATESTCREATED         LATESTREADY           READY   REASON
3
helloworld      https://helloworld.default.knative.chi.coreweave.com      helloworld-ngzsn      helloworld-ngzsn      True
Copied!
Monitoring
Managed Grafana provides monitoring of requests, success rates, response times and auto-scaling metrics transparently. No metrics specific code needs to be added to the serverless application.

Bring Your Own IP

Next - CoreWeave Kubernetes

Metrics

Last modified 2mo ago

Copy link

Contents

Monitoring