Set Up High Availability

🕒 15 min read 🔄 Updated: Oct 24, 2025 🏷️ Architecture

High availability (HA) ensures your applications remain operational during hardware failures, network outages, or maintenance windows. CloudNexus provides native tools to deploy active-active or active-standby architectures across multiple regions with automated failover, zero-downtime updates, and consistent global latency.

This guide walks you through provisioning a production-ready HA setup using the CloudNexus CLI, managed load balancers, replicated databases, and intelligent health checking.

Prerequisites

CloudNexus account with Professional or Enterprise tier
CloudNexus CLI installed and authenticated: cx-cli auth login
Access to at least two geographic regions
Application containerized (Docker) or compatible with CloudNexus App Server

Target Architecture

The recommended HA topology uses a global load balancer distributing traffic to regional clusters. Each region runs independently synchronized instances, with an asynchronous primary-replica database setup for low-latency reads and crash-safe writes.

🌐Global LB

🔍Health Checks

🔒SSL/TLS

🇺🇸US-East Cluster

🇪🇺EU-West Cluster

🇯🇵AP-Tokyo Cluster

🗄️Primary DB

🔄Async Replicas

📊Metrics

Step-by-Step Setup

Step 1

Provision Multi-Region Nodes

Create identical compute clusters across your target regions. Use the cx-cli to define node pools with identical resource allocations.

                            bash
                            
# Create regional node pools
cx-cli cluster create --region us-east-1 --name ha-cluster-us --nodes 3
cx-cli cluster create --region eu-west-1 --name ha-cluster-eu --nodes 3
cx-cli cluster create --region ap-tokyo-1 --name ha-cluster-ap --nodes 2

# Verify cluster status
cx-cli cluster list --status running

Step 2

Configure Global Load Balancer

Deploy an anycast load balancer that routes traffic based on latency, availability, and custom health probes. Enable sticky sessions if your application requires state persistence.

yaml

# lb-config.yaml
loadbalancer:
  name: global-ha-lb
  type: anycast
  algorithm: least-connections
  sticky_session: true
  tls:
    cert_arn: arn:cx:acm:us-east-1:cert-48291
  targets:
    - cluster: ha-cluster-us
      weight: 40
    - cluster: ha-cluster-eu
      weight: 35
    - cluster: ha-cluster-ap
      weight: 25

💡

Tip Use --health-path /healthz when applying the config to ensure the LB only routes to healthy nodes.

bash

cx-cli lb create --config lb-config.yaml --health-path /healthz

Step 3

Set Up Managed Database Replication

Create a primary database instance in your main region and configure asynchronous read replicas in secondary regions. This minimizes write latency while providing failover-ready data.

                            bash
                            
# Create primary PostgreSQL cluster
cx-cli db create --engine postgresql --tier ha-primary --region us-east-1 --name app-db-primary

# Add cross-region async replicas
cx-cli db replica create --source app-db-primary --region eu-west-1 --mode async
cx-cli db replica create --source app-db-primary --region ap-tokyo-1 --mode async

Step 4

Enable Automated Failover & Health Monitoring

Activate the CloudNexus Health Gateway to continuously probe all endpoints. Configure automatic promotion rules so replicas become primary within seconds of a region failure.

bash

cx-cli failover configure --cluster-group ha-clusters \\
  --health-interval 5s \\
  --promote-on region-down,health-check-fail \\
  --dns-failover enabled

⚠️

Important DNS-based failover depends on TTL settings. Ensure your DNS provider is set to ≤60 seconds for rapid propagation during outages.

Verification & Testing

Run the following commands to validate your HA setup before going live:

Health Check Validation: cx-cli health run --all --verbose
Failover Simulation: cx-cli failover simulate --target us-east-1
Latency Map: cx-cli lb trace --global

Expected output shows all regions in HEALTHY state, with failover time < 3.2s and zero packet loss during region simulation.

Best Practices

Idempotent Deployments: Ensure your app handles duplicate requests gracefully during failover windows.
Connection Pooling: Use PgBouncer or ProxySQL to manage database connections across regions.
Backup Strategy: Enable cross-region automated snapshots with --retention 30d.
Cost Optimization: Use traffic shaping to route non-critical workloads to cheaper regions during peak hours.
Monitoring: Integrate with CloudNexus Metrics API or Prometheus for real-time SLA tracking.

← Auto Scaling Guide Database Clusters →