Scaling up your application in order to handle workload

Prerequisite

Manual Scale Application

  • click topology in left menu, click Duke icon (backend deployment), Details tab
  • click increase ther pod count (^ icon) to 2 Pod
  • wait until application scale to 2 Pods (circle around Duke icon change to dark blue)
  • Test load to application, go to web terminal, run below command 
    BACKEND_URL=https://$(oc get route backend -o jsonpath='{.spec.host}')
while [  1  ];
do
  curl $BACKEND_URL/backend
  printf "\n"
  sleep 10
done
    example result, check have result from 2 pods (Host value)
    Backend version:v1, Response:200, Host:backend-95647fbb8-kt886, Status:200, Message: Hello, World
Backend version:v1, Response:200, Host:backend-95647fbb8-q9dqv, Status:200, Message: Hello, World
Backend version:v1, Response:200, Host:backend-95647fbb8-kt886, Status:200, Message: Hello, World
Backend version:v1, Response:200, Host:backend-95647fbb8-q9dqv, Status:200, Message: Hello, World
  • after few minute, type 'ctrl-c' in web terminal to terminated curl command
  • go to Resources Tab, in Pods section, show 2 pods after scale
  • click 'View logs' of 1st Pod and 2nd Pod to confirm both pod are processed. example of 1st pod
    example of 2nd pod
  • back to detail pages of backend deployment, scale pod to 0 (for this case, no pod for this application)
  • scale backend to 1 pod

Auto Scale Application

  • add HorizontalPodAutoscaler
  • go to Topology, click at Duke icon for open backend deployment, click action dropdown menu, select Add HorizontalPodAutoscaler
  • in Add HorizontalPodAutoscaler, use Form view
    • set Name: example
    • Minimum Pods: 1
    • Maximum Pods: 3
    • CPU Utilization: 1%
  • click save, and wait until backend deployment change to Autoscaling
  • load test to backend application for proof auto scale
  • go to web terminal
  • run load test command
    BACKEND_URL=https://$(oc get route backend -o jsonpath='{.spec.host}')
while [  1  ];
do
  curl $BACKEND_URL/backend
  printf "\n"
done
  • click detail tab of backend deployment, wait until autoscaled to 3 (wait a few minutes)
  • click resources tab, see 3 pods auto scale
  • click Observe tab to view CPU usage
  • back to web terminal, input 'ctrl-c' to terminate load test command
  • wait 5 minute, autoscaled will reduce pod to 1.
  • remove HorizontalPodAutoscaler, go to backend deployment information page, select action menu, select remove HorizontalPodAutoscaler
  • confirm Remove, and wait until backend change to manual scale
  • Optional: if you don't want to wait autoscale down to 1 pod, you can remove HorizontalPodAutoscaler and manual scale down to 1 by yourself.

Next Step

results matching ""

    No results matching ""