Redundancy and scaling

Because we rely on a small number of queues, introducing redundancy and scalability becomes trivial.

We build each of those services into a docker container that's deployed on Kubernetes (GKE) with multiple pods per deployment. Each one is configured to scale in accordance with the depth of the queue.

apiVersion: autoscaling/v2beta2
kind: HorizontalPodAutoscaler
metadata:
  name: pubsub
spec:
  minReplicas: 2
  maxReplicas: 20
  metrics:
  - external:
      metric:
       name: pubsub.googleapis.com|subscription|num_undelivered_messages
       selector:
         matchLabels:
           resource.labels.subscription_id: SUBSCRIPTION_ID
      target:
        type: AverageValue
        averageValue: 0
    type: External
  scaleTargetRef:
    apiVersion: apps/v1
    kind: Deployment
    name: delivery-workers

How We Built a Rate Limiter for Outbound Webhooks

Instantly & reliably manageyour webhooks

Instantly & reliably manage
your webhooks