Hybrid Cloud Bursting Architecture

Automatically scale beyond on-premise capacity with secure AWS bursting during traffic spikes. This production-ready solution combines Terraform for cloud infrastructure and Ansible for configuration management.

Features

• Auto-Scaling: Burst to AWS when on-prem CPU >70% for 5 minutes
• High Availability: <3s failover with HAProxy + Keepalived VIP
• Secure Connectivity: WireGuard VPN with PSK authentication
• Database Resilience: MySQL Galera multi-master cluster
• Cost Monitoring: Built-in AWS budget alerts

Architecture Components

On-Premise

• HAProxy: Load balancing with dynamic backend updates
• Keepalived: Virtual IP (192.168.1.200) for failover
• MySQL Galera: Synchronous multi-master replication
• WireGuard: Secure tunnel to AWS (UDP 51820)

AWS Cloud

• Auto Scaling Group: t3.medium spot instances
• VPC Networking: Isolated 10.0.0.0/16 with VPN peering
• CloudWatch Alarms: CPU/Memory monitoring

Deployment

Prerequisites

• 2+ Ubuntu 22.04 servers (4vCPU/8GB RAM each)
• AWS account with EC2/VPC permissions
• Terraform 1.5+ and Ansible 8+

Installation Steps

1. Clone the repository:

git clone https://github.com/yourrepo/hybrid-cloud-bursting.git
cd hybrid-cloud-bursting

2. Initialize Terraform:

cd terraform
terraform init

3. Deploy AWS resources (edit terraform.tfvars first):

terraform apply -var-file=production.tfvars

4. Configure on-premise servers:

ansible-playbook -i ansible/inventory/on_prem.yml ansible/playbooks/setup.yml

Configuration Files

File	Purpose
terraform/modules/autoscaling/main.tf	AWS scaling policies
ansible/playbooks/haproxy.yml	Dynamic load balancer config
configs/haproxy/haproxy.cfg	Load balancing rules
scripts/wg-install.sh	WireGuard VPN setup

Monitoring

Access these dashboards after deployment:

• Grafana: http://<on-prem-ip>:3000
  • Default credentials: admin/grafana-admin
• Prometheus: http://<on-prem-ip>:9090

Key metrics tracked:

• haproxy_backend_http_requests_total
• mysql_global_status_threads_connected
• aws_ec2_cpuutilization_average

Testing Procedures

Load Testing

locust -f load_testing/burst_test.py --users 1000 --spawn-rate 50

Failover Test

# On primary node:
sudo systemctl stop keepalived
# Verify VIP migrates to backup within 3 seconds

Cost Management

Resource	Estimated Cost
AWS Burst Instances	$0.08/hr (spot)
VPN Data Transfer	$0.09/GB
Monitoring	$0.30/day

Set budget alerts in AWS when monthly spend exceeds $500.

Troubleshooting

Common Issues:

1. VIP Not Failing Over:

   • Check Keepalived logs: journalctl -u keepalived
   • Verify ARP tables on network switches

2. Database Split-Brain:

   SHOW STATUS LIKE 'wsrep%';
   # Re-bootstrap cluster if needed
   sudo galera_new_cluster

3. AWS Instances Not Scaling:

   • Verify CloudWatch alarms exist
   • Check Auto Scaling Group health checks

License

Apache License 2.0 - See LICENSE for full text.