Integrating Cloud Load Balancer with Container Orchestration for High Availability

Abstract

The rapid adoption of cloud-native architectures, driven by containerization and orchestration platforms such as Kubernetes, has transformed the scalability and flexibility of modern applications. Yet, achieving true high availability (HA) in such dynamic environments remains challenging, as traditional load balancing approaches—such as Round Robin or Least Connections—are inherently reactive and fail to capture the transient health conditions of containerized workloads. This paper presents an integrated approach to cloud load balancing and container orchestration for high availability through the Predictive and Health-Aware Load Balancing (PHAL) algorithm. PHAL combines two synergistic components: a Predictive Forecasting Module (PFM), which leverages Long Short-Term Memory (LSTM) networks to anticipate workload surges and proactively scale resources, and a Health-Aware Routing Module (HRM), which dynamically distributes traffic across pods based on a composite health score derived from latency, CPU utilization, active connections, and predicted headroom. Through experimental evaluation on a Google Kubernetes Engine cluster running a multi-service e-commerce application, PHAL demonstrates significant improvements over conventional load balancing strategies, including reduced tail latency, enhanced throughput, and resilience against node and pod failures. The study highlights that integrating intelligent cloud load balancing with orchestration frameworks is essential for ensuring robust, fault-tolerant, and highly available cloud-native systems.