cepdnaclk/e19-4yp-AI-Dirven-Latency-Constrained-Resource-Management-In-Kubernetes

Understanding the Problem Statement: Kubernetes Resource Auto-Updater with Latency Monitoring

Overview

This setup is designed to analyze latency increments during dynamic resource updates (CPU/Memory) in a Kubernetes environment. It uses Apache JMeter to simulate traffic and measure latency, and Prometheus to collect metrics. An Auto Updater module adjusts resource allocations based on historical performance data.

Architecture

Components

• Clients (Local Environment):
  Use Apache JMeter to generate HTTP traffic towards services in the Kubernetes cluster. Latency logs are collected and stored for analysis.

• Apache JMeter:
  Load testing tool used to simulate concurrent requests and log latency data for each request.

• Kubernetes Cluster:

  • Services (Service 1, Service 2): Sample applications that process HTTP requests.

  • Prometheus: Collects performance metrics (CPU, memory, etc.) from services.

  • Auto Updater: Uses 1 hour of historical data from Prometheus to compute new resource requirements.

  • Kubernetes API: Applies resource updates (requests/limits) sent by the Auto Updater.

Objectives

• Measure latency impact during live resource updates.

• Evaluate whether resource adjustments introduce unacceptable latency spikes.

• Optimize auto-scaling strategies (vertical scaling focus).

Prerequisites

• Kubernetes Cluster (Minikube, Kind, GKE, etc.)

• Apache JMeter

• Prometheus installed in the cluster

• Python or Go-based Auto Updater (custom module)

• Grafana (optional, for metric visualization)

Setup Steps

1. Deploy Prometheus

• Follow the Prometheus Helm chart to install Prometheus in your cluster.

2. Deploy Sample Services

• Deploy at least two services that handle HTTP requests.

3. Configure JMeter

• Create a test plan to send concurrent HTTP requests to the services.
• Enable latency logging and store results locally.

4. Run Auto Updater

• Continuously fetch metrics from Prometheus.
• Based on predefined rules or models, send resource update requests to the Kubernetes API.