0010: AI4Canvas Operator Scope Decision¶
Status¶
Proposed
Context¶
A proposal has been made to create an AI4Canvas Operator that would:
Predict ODA Component workload patterns
Make cognitive recommendations for resource allocation
Provide goal-driven resource optimization
Enable continuous optimization through machine learning
Decision¶
We propose to not create a TMF-specific AI4Canvas Operator because:
Existing Kubernetes Solutions:
Kubernetes already has mature autoscaling capabilities (HPA, VPA, Cluster Autoscaler)
Projects like Keda provide event-driven autoscaling
Kubernetes Event-driven Autoscaling (KEDA) supports multiple scalers and metrics
Tools like Prometheus Operator handle monitoring
Industry Standards:
Resource optimization is not telco-specific
Following cloud-native practices promotes interoperability
Many cloud providers offer AI-driven optimization tools:
AWS Auto Scaling
Google Cloud’s Autopilot
Azure’s AKS auto-scaling
Scope Consideration:
TMF should focus on telco-specific aspects
Resource optimization is a general cloud computing concern
Reinventing existing solutions could lead to fragmentation
Consequences¶
Positive¶
Leverage battle-tested Kubernetes ecosystem
Reduce maintenance burden
Benefit from broader community support
Stay aligned with cloud-native practices
Negative¶
May require integration work to connect existing tools
Less direct control over optimization algorithms
Need to adapt to Kubernetes-native approaches
Recommendation¶
Instead of creating a new operator, we recommend:
Document best practices for using existing Kubernetes autoscaling
Create guidelines for integrating with cloud provider optimization tools
Focus TMF efforts on telco-specific aspects of ODA Components