DRL Agent in Concurrent Service Auto-scaling for Knative Resource quota-based Serverless System paper

https://doi.org/10.1016/j.future.2024.06.019

-input: ML Traffic Prediction of all concurrent services

-output: DRLScaleAction CRD (the Quota-based Knative Hybrid autoscaling Operator uses this as input to apply the scaling action)

Requirements for running

• Install DRLScaleAction CRD from https://github.com/mipearlska/knative_drl_operator
• Install requirements (Python >= 3.6.9)

sudo apt install python3-dev python3-venv libffi-dev gcc libssl-dev git

python3 -m venv $HOME/DRL_knative_scale_limit_resources

source $HOME/DRL_knative_scale_limit_resources/bin/activate

pip install -U pip

pip install setuptools==65.5.0 pip==21

pip install wheel==0.38.0

• Then

pip install -r requirements.txt

• Then

pip install stable-baselines3

pip install sb3-contrib

• Locust Traffic Generator (https://locust.io/). Recommended install version 2.8.6.

pip3 install locust==2.8.6

• Locust exporter for Prometheus. Locust exporter cannot fetch 95%,50% latency for some recent versions of Locust (> 3).

sudo docker run -d --net=host containersol/locust_exporter

• When deploying Prometheus, add Locust config in values.yaml of Prometheus: job_name = Locust's job parameter in Prometheus query below target = "Host_URL_that_run_Locust:9646"

prometheus:
  prometheusSpec:
    additionalScrapeConfigs:
      - job_name: "generator1"
        scrape_interval: 2s
        static_configs:
        - targets: ["192.168.26.42:9646"]
      - job_name: "generator2"
        scrape_interval: 2s
        static_configs:
        - targets: ["192.168.26.20:9646"]
      - job_name: "generator3"
        scrape_interval: 2s
        static_configs:
        - targets: ["192.168.26.41:9646"]  

• Modify a correct Prometheus URL,port in configs/configs.py

PROMETHEUS_URL = "http://192.168.26.42:32000"

• Provide correct Prometheus IP and job_name in get_traffic_metric() function inside def predict(api, model) of the main.py file:

    get_traffic_metric("192.168.26.42", "generator1", "house")
    get_traffic_metric("192.168.26.20", "generator2", "senti")
    get_traffic_metric("192.168.26.41", "generator3", "numbr")

• (Optional) Modify when (which second) to make a prediction in main function of main.py file:

schedule.every().minute.at(":55").do(lambda: predict(api))

Running

• Generate Traffic to service by running locust traffic profile (Given sample profile change traffic amount every 1 minute)

locust -f locustservicetraffic.py

• Run Traffic Prediction service

python main.py