Skip to content

deployment.md

Latest commit

 

History

History
108 lines (79 loc) · 3.45 KB

deployment.md

File metadata and controls

108 lines (79 loc) · 3.45 KB

GKE Deployment & Performance Profiling

This guide covers the steps for deploying the DeathStarBench Social Network to Google Kubernetes Engine (GKE) and performing performance profiling. It includes steps from setting up the project in GKE to deploying the application, simulating workloads, and analyzing performance.

Table of Contents

  1. Prerequisites
  2. Setting up GKE Cluster
  3. Deploying the Application Using Helm
  4. Profiling CPU, Memory, and Latency
  5. Scaling and Load Balancing
  6. Troubleshooting

Prerequisites

  1. Google Cloud Project

    • Set up a Google Cloud Project and enable Kubernetes Engine.
    • Ensure gcloud CLI is installed and authenticated with your Google Cloud account.

  2. Google Kubernetes Engine (GKE)

    • Install kubectl and gcloud CLI tools.
    • Enable Kubernetes Engine API in your project.
    • Set your project ID and configure the gcloud CLI.

  3. Helm

    • Install Helm for managing Kubernetes applications.

Setting up GKE Cluster

  1. Create a Kubernetes cluster in GKE:

    gcloud container clusters create deathstar-cluster \
  --zone=asia-south1-a \
  --cluster-version=latest \
  --num-nodes=1

  2. Configure kubectl to use the new cluster:

    gcloud container clusters get-credentials deathstar-cluster --zone asia-south1-a

  3. Verify the cluster status:

    kubectl get nodes

Deploying the Application Using Helm

  1. Clone the DeathStarBench repository and navigate to the socialNetwork folder:

    git clone https://github.com/DeathStarBench/DeathStarBench.git
cd DeathStarBench/socialNetwork/helm-chart/socialnetwork

  2. Create a namespace for the application:

    kubectl create namespace deathstarbench

  3. Install the application using Helm:

    helm install social-network . -n deathstarbench --create-namespace

  4. Check the status of the pods:

    kubectl get pods -n deathstarbench

Profiling CPU, Memory, and Latency

To profile the application's performance:

  1. Simulate Workload:

    • Write a script to simulate a read and write workload for the social network.

  2. Use Prometheus and Grafana:

    • Install Prometheus for monitoring:

      helm install prometheus stable/prometheus --namespace deathstarbench

    • Set up Grafana for visualizing performance metrics.

    • Configure Prometheus as a data source in Grafana.

  3. Run the Workload and Monitor:

    • Start the workload script and monitor the CPU, memory usage, and latency using Grafana.

Scaling and Load Balancing

To scale the deployment and improve load balancing:

  1. Horizontal Pod Autoscaling:

    • Configure Horizontal Pod Autoscaler (HPA) to automatically scale the number of replicas based on CPU or memory usage.

  2. Cluster Autoscaler:

    • Enable Google Cloud's Cluster Autoscaler to scale the cluster nodes based on workload demands.

  3. Test Different Configurations:

    • Test different configurations by adjusting the number of nodes or using multiple node pools.

Troubleshooting

If you encounter issues such as pod scheduling errors due to insufficient resources, consider the following:

  • Check the current node resources: 
    kubectl describe node <node-name>