Autoscale Sample App - Go

A demonstration of the autoscaling capabilities of a Knative Serving Revision.

Prerequisites

  1. A Kubernetes cluster with Knative Serving installed.

  2. A metrics installation for viewing scaling graphs (optional).

  3. The hey load generator installed (go get -u github.com/rakyll/hey).

  4. Clone this repository, and move into the sample directory:

    git clone -b "release-0.14" https://github.com/knative/docs knative-docs
cd knative-docs

Deploy the Service

  1. Deploy the sample Knative Service:

    kubectl apply --filename docs/serving/samples/autoscale-go/service.yaml

  2. Obtain the URL of the service (once Ready):

    $ kubectl get ksvc autoscale-go
NAME            URL                                                LATESTCREATED         LATESTREADY           READY   REASON
autoscale-go    http://autoscale-go.default.1.2.3.4.xip.io    autoscale-go-96dtk    autoscale-go-96dtk    True

Load the Service

  1. Make a request to the autoscale app to see it consume some resources.

    curl "http://autoscale-go.default.1.2.3.4.xip.io?sleep=100&prime=10000&bloat=5"

    Allocated 5 Mb of memory.
The largest prime less than 10000 is 9973.
Slept for 100.13 milliseconds.

  2. Send 30 seconds of traffic maintaining 50 in-flight requests.

    hey -z 30s -c 50 \
  "http://autoscale-go.default.1.2.3.4.xip.io?sleep=100&prime=10000&bloat=5" \
  && kubectl get pods

    Summary:
  Total:        30.3379 secs
  Slowest:      0.7433 secs
  Fastest:      0.1672 secs
  Average:      0.2778 secs
  Requests/sec: 178.7861

  Total data:   542038 bytes
  Size/request: 99 bytes

Response time histogram:
  0.167 [1]     |
  0.225 [1462]  |■■■■■■■■■■■■■■■■■■■■■■■■■■■■■■■
  0.282 [1303]  |■■■■■■■■■■■■■■■■■■■■■■■■■■■■
  0.340 [1894]  |■■■■■■■■■■■■■■■■■■■■■■■■■■■■■■■■■■■■■■■■
  0.398 [471]   |■■■■■■■■■■
  0.455 [159]   |■■■
  0.513 [68]    |■
  0.570 [18]    |
  0.628 [14]    |
  0.686 [21]    |
  0.743 [13]    |

Latency distribution:
  10% in 0.1805 secs
  25% in 0.2197 secs
  50% in 0.2801 secs
  75% in 0.3129 secs
  90% in 0.3596 secs
  95% in 0.4020 secs
  99% in 0.5457 secs

Details (average, fastest, slowest):
  DNS+dialup:   0.0007 secs, 0.1672 secs, 0.7433 secs
  DNS-lookup:   0.0000 secs, 0.0000 secs, 0.0000 secs
  req write:    0.0001 secs, 0.0000 secs, 0.0045 secs
  resp wait:    0.2766 secs, 0.1669 secs, 0.6633 secs
  resp read:    0.0002 secs, 0.0000 secs, 0.0065 secs

Status code distribution:
  [200] 5424 responses

    NAME                                             READY   STATUS    RESTARTS   AGE
autoscale-go-00001-deployment-78cdc67bf4-2w4sk   3/3     Running   0          26s
autoscale-go-00001-deployment-78cdc67bf4-dd2zb   3/3     Running   0          24s
autoscale-go-00001-deployment-78cdc67bf4-pg55p   3/3     Running   0          18s
autoscale-go-00001-deployment-78cdc67bf4-q8bf9   3/3     Running   0          1m
autoscale-go-00001-deployment-78cdc67bf4-thjbq   3/3     Running   0          26s

Analysis

Algorithm

Knative Serving autoscaling is based on the average number of in-flight requests per pod (concurrency). The system has a default target concurrency of 100 but we used 10 for our service. We loaded the service with 50 concurrent requests so the autoscaler created 5 pods (50 concurrent requests / target of 10 = 5 pods)

Panic

The autoscaler calculates average concurrency over a 60 second window so it takes a minute for the system to stablize at the desired level of concurrency. However the autoscaler also calculates a 6 second panic window and will enter panic mode if that window reached 2x the target concurrency. In panic mode the autoscaler operates on the shorter, more sensitive panic window. Once the panic conditions are no longer met for 60 seconds, the autoscaler will return to the initial 60 second stable window.

                                                       |
                                  Panic Target--->  +--| 20
                                                    |  |
                                                    | <------Panic Window
                                                    |  |
       Stable Target--->  +-------------------------|--| 10   CONCURRENCY
                          |                         |  |
                          |                      <-----------Stable Window
                          |                         |  |
--------------------------+-------------------------+--+ 0
120                       60                           0
                     TIME

Customization

The autoscaler supports customization through annotations. There are two autoscaler classes built into Knative:

  1. kpa.autoscaling.knative.dev which is the concurrency-based autoscaler described above (the default), and

  2. hpa.autoscaling.knative.dev which delegates to the Kubernetes HPA which autoscales on CPU usage.

    Example of a Service scaled on CPU:

    apiVersion: serving.knative.dev/v1
kind: Service
metadata:
  name: autoscale-go
  namespace: default
spec:
  template:
    metadata:
      annotations:
        # Standard Kubernetes CPU-based autoscaling.
        autoscaling.knative.dev/class: hpa.autoscaling.knative.dev
        autoscaling.knative.dev/metric: cpu
    spec:
      containers:
        - image: gcr.io/knative-samples/autoscale-go:0.1

    Additionally the autoscaler targets and scaling bounds can be specified in annotations. Example of a Service with custom targets and scale bounds:

    apiVersion: serving.knative.dev/v1
kind: Service
metadata:
  name: autoscale-go
  namespace: default
spec:
  template:
    metadata:
      annotations:
        # Knative concurrency-based autoscaling (default).
        autoscaling.knative.dev/class: kpa.autoscaling.knative.dev
        autoscaling.knative.dev/metric: concurrency
        # Target 10 requests in-flight per pod.
        autoscaling.knative.dev/target: "10"
        # Disable scale to zero with a minScale of 1.
        autoscaling.knative.dev/minScale: "1"
        # Limit scaling to 100 pods.
        autoscaling.knative.dev/maxScale: "100"
    spec:
      containers:
        - image: gcr.io/knative-samples/autoscale-go:0.1

    Note: for an hpa.autoscaling.knative.dev class service, the autoscaling.knative.dev/target specifies the CPU percentage target (default "80").

Demo

View the Kubecon Demo of Knative autoscaler customization (32 minutes).

Dashboards

View the Knative Serving Scaling and Request dashboards (if configured).

kubectl port-forward --namespace knative-monitoring $(kubectl get pods --namespace knative-monitoring --selector=app=grafana  --output=jsonpath="{.items..metadata.name}") 3000

scale dashboard

request dashboard

Other Experiments

  1. Send 60 seconds of traffic maintaining 100 concurrent requests.

    hey -z 60s -c 100 \
  "http://autoscale-go.default.1.2.3.4.xip.io?sleep=100&prime=10000&bloat=5"

  2. Send 60 seconds of traffic maintaining 100 qps with short requests (10 ms).

    hey -z 60s -q 100 \
  "http://autoscale-go.default.1.2.3.4.xip.io?sleep=10"

  3. Send 60 seconds of traffic maintaining 100 qps with long requests (1 sec).

    hey -z 60s -q 100 \
  "http://autoscale-go.default.1.2.3.4.xip.io?sleep=1000"

  4. Send 60 seconds of traffic with heavy CPU usage (~1 cpu/sec/request, total 100 cpus).

    hey -z 60s -q 100 \
  "http://autoscale-go.default.1.2.3.4.xip.io?prime=40000000"

  5. Send 60 seconds of traffic with heavy memory usage (1 gb/request, total 5 gb).

    hey -z 60s -c 5 \
  "http://autoscale-go.default.1.2.3.4.xip.io?bloat=1000"

Cleanup

kubectl delete --filename docs/serving/samples/autoscale-go/service.yaml

Further reading

Autoscaling Developer Documentation