Skip to main content
Uncategorized

Cross-Region Centralized Backup & Restore Implementation

By June 29, 2026July 3rd, 2026No Comments

 

Cross-Region Centralized Backup & Restore Implementation

 

1. Project Overview

  • AWS Customer Name: Pinfi
  • AWS Partner Name: Techpartner
  • DR Strategy: Backup & Restore
  • Industry: FinTech / Payment Acquirer
  • Workload: Mission-Critical Payment Processing & Merchant Onboarding

2. Customer Challenge

Pinfi operated their core payment platform in a single AWS region (ap-south-1 Mumbai). As a critical payment acquirer for partners like Jiobank and Finobank, any regional outage posed a catastrophic risk to merchant operations and financial settlement.

Specific Challenges:

  • Single Region Dependency: An outage in Mumbai would halt global payment operations, leading to critical revenue loss and breach of banking partner SLAs.
  • Strict Financial RPO: Regulatory and reconciliation mandates required a 30-minute Recovery Point Objective (RPO).
  • Banking Handshake Continuity: Banking partners utilize IP whitelisting; recovery requires a strategy to maintain whitelisted connectivity without manual re-configuration delays.
  • Cost Optimization: The business required a resilient solution, but prioritized cost-optimization over instantaneous failover, seeking to minimize idle infrastructure costs.

3. Partner Solution

Techpartner designed and implemented a Cross-Region Backup architecture. This strategy was selected as the most cost-effective path to meeting 30-minute RPO and 4-hour RTO targets by utilizing a ‘Cold DR’ model in Hyderabad (ap-south-2).

Technical Architecture & Implementation

The solution deliberately avoids third-party or overarching backup services like AWS Backup to maintain clean cost attribution and simplified operational ownership. Instead, it relies entirely on AWS native automation primitives.

A. Cross-Region Data Persistence

Techpartner implemented independent protection for the following resource types:

  • Amazon RDS (MySQL 8.0): Configured with RDS Native Automated Backups featuring cross-region copy to Hyderabad, enabling Point-in-Time Recovery (PITR) to meet the 30-minute RPO.
  • Amazon EC2: Amazon Data Lifecycle Manager (DLM) policies automate hourly AMI copies to the secondary region, ensuring the OS and application state for the Java transaction engines, Laravel portal, and logging server are always available.
  • S3 Buckets: Asynchronous Cross-Region Replication (CRR) for merchant documentation, SFTP logs, and deployment artifacts.

B. Automated Restoration via Terraform & CodeDeploy

To achieve the 4-hour RTO, Techpartner developed a modular Terraform library that automates the provisioning of the Recovery VPC in Hyderabad. This library programmatically identifies the latest DLM AMI copy and initiates the PITR restoration of RDS. The stack includes:

  • VPC Networking and Subnets.
  • Pre-allocated and Whitelisted Elastic IPs (EIPs) for banking handshakes.
  • Security Groups and IAM Roles.
  • Application Load Balancers (ALB).

Following infrastructure hydration, an independent AWS CodePipeline—provisioned exclusively by Terraform in Hyderabad—triggers CodeDeploy to pull the latest artifacts from the replicated S3 bucket, restoring the Java and Laravel applications without relying on the primary region’s CI/CD plane.

4. High-Level and Technical Detail of AWS Services Used

  • Amazon RDS & Amazon DLM: Native orchestration engines for cross-region data replication and AMI snapshot copying.
  • Terraform (IaC): Primary tool for automated regional infrastructure hydration, ensuring 100% environment parity.
  • AWS CodePipeline & CodeDeploy: Fully decoupled CI/CD deployment mechanisms running independently in the recovery region.
  • AWS Lambda & AWS Step Functions: Powering a custom observability state machine that verifies backup copy age every 30 minutes and triggers out-of-band SNS alerts on failure.

5. Outcomes and Quantitative Results

The implementation was validated through a formal DR Game Day on Jun 17, 2026, simulating a total regional failure in Mumbai.

Quantitative Metrics:

  • RPO Achievement: Verified 30-minute data freshness in Hyderabad using RDS PITR logs.
  • RTO Achievement: Full platform restoration achieved in 3 hours and 55 minutes (Target: 4 hours).
  • Cost Efficiency: Saved approximately $900/month by utilizing a Backup & Restore model over a Pilot Light or Warm Standby approach.
  • Operational Confidence: Successfully restored banking connectivity via pre-whitelisted EIPs without manual intervention.

6. Conclusion

By leveraging AWS’s native backup orchestration and Terraform-led automation, Techpartner successfully transitioned Pinfi to a resilient, audit-ready posture. The Cross-Region Backup strategy provides the perfect balance of cost-optimization and rigorous recovery capabilities, ensuring Pinfi can maintain merchant operations during a total regional failure.