Carbon Aware Scaling Poc

This project is to integrate CO2 emission intensity data into KEDA scaling decision.

Overview

The background, architecture, findings and demo can be found on the YouTube playlist

Data Sources

• National Grid ESO https://data.nationalgrideso.com/search?q=carbon&sort=score%3Adesc
• Wattime https://www.watttime.org/
• Electricity Maps https://static.electricitymaps.com/api/docs/index.html

Deployment Scheme

Relates to kedacore/keda#3467

Target User Experience:

• Admin can enable Carbon awareness for kubernetes resources, without requiring applications code change.
• Admin defines Carbon Intensity thresholds to scale up/down based on, and AllowedMaxReplicaCount as a scaling target for the Scaled Object (Deployment or StatefulSet).

apiVersion: keda.sh/v1alpha1
kind: ScaledObject
metadata:
  name: {scaled-object-name}
spec:
  scaleTargetRef:
    name:          {name-of-target-resource}         # Mandatory. Must be in the same namespace as the ScaledObject
  maxReplicaCount:  100                              # Optional. Default: 100
  environmentalImpact:
    carbon:
    - measuredEmission: 300
      allowedMaxReplicaCount: 50
    - measuredEmission: 500
      allowedMaxReplicaCount: 10
  fallback:                                          # Optional. Section to specify fallback options
    failureThreshold: 3                              # Mandatory if fallback section is included
    replicas: 6                                      # Mandatory if fallback section is included
  triggers:
  # {list of triggers to activate scaling of the target resource}

Required data

• Customer input: define for the ScaledObject, the desired values for measuredEmission and allowedReplicaCount
• Platform Metrics: Location based Carbon Marginal Intensity (also called Electricity Carbon Intensity), expressed in gCO2eq/Kwh (Location based means it depends on the country / Region / Data Center where the K8s cluster is deployed)

Engineering solution

• for POC: start with python
• Frameworks
  • Kubernetes client / sdk
  • Kopf: Kubernetes operator framework --> faciliates writing k8s operators: hooking to k8s resource CRUD + Events and defining handlers for them
    • https://kopf.readthedocs.io/en/stable/timers/

Architecture of poc vo

• Operator code: handlers.py + DockerFile: Timer, regularly checks Carbon Intensity and scales up / down, Keda Scaled Object
• Operator deploy: OperatorDeployment.yaml + rbac.yaml
• Operator target resource: Keda Http scaled object
• Keda scaled Object Target : Dummy Deployment

Open Items

• Determine initial data provider to integrate
• Integrate Carbon Intensity Data via K8s Metrics (Prometheus)
• Update POC to use real-world metrics rather than mock
• Define proposal for KEDA around Carbon Aware Scaling and how to integrate it as an extension