Load-Balancing of Kubernetes-Based Edge Computing Infrastructure Using Resource Adaptive Proxy

Area of Science:

• Edge Computing
• Container Orchestration
• Network Load Balancing

Background:

• Kubernetes (K8s) is a key container orchestration tool for edge computing.
• K8s default load balancing (kube-proxy) can cause significant delays due to geographically dispersed nodes and worker overload.
• Existing mechanisms lack adaptive resource awareness for optimal request distribution.

Purpose of the Study:

• To propose an enhanced load balancer, Resource Adaptive Proxy (RAP), for Kubernetes in edge computing environments.
• To address the limitations of default K8s load balancing, specifically request latency and throughput.
• To improve the efficiency and performance of containerized applications at the edge.

Main Methods:

• Developed Resource Adaptive Proxy (RAP) to monitor pod resource status and network conditions.
• Implemented intelligent load-balancing decisions prioritizing local handling and adaptive forwarding.
• Compared RAP's performance against K8s default load balancing through experimental evaluation.

Main Results:

• RAP significantly improved throughput compared to the default K8s load balancing mechanism.
• RAP demonstrated a substantial reduction in request latency, especially in edge scenarios.
• Experimental results validated RAP's effectiveness in dynamic edge environments.

Conclusions:

• Resource Adaptive Proxy (RAP) offers a superior load-balancing solution for Kubernetes in edge computing.
• RAP's adaptive resource monitoring and intelligent routing effectively mitigate latency and enhance performance.
• The proposed method provides a scalable and efficient approach for edge infrastructure management.