Migrating to Azure for ERP Reliability and Scale

We migrated the Varuni demo to Azure to standardize environments, improve reliability, and simplify support. Here is the blueprint we used and what we learned.

Goals

• Consistent environments from dev → staging → prod.
• Secure networking and secret management by default.
• Predictable autoscaling and cost controls.
• Observability that shortens time-to-diagnose.

Baseline architecture

• Containerized services with image promotion via tags.
• Azure Container Apps or AKS depending on scale/ops preferences.
• Azure Key Vault for secrets; Managed Identity for service-to-service auth.
• Azure Monitor + Log Analytics for metrics, traces, and logs.

Environment strategy

• Immutable builds; environment configured via parameters and secrets.
• Feature flags for risk-managed rollout.
• Staging mirrors production topology (reduced size).

Networking and security

• Private endpoints for databases and storage.
• Ingress controlled with Web Application Firewall (WAF).
• Egress allow-listing for third-party APIs (e.g., POS providers).

Observability

• Structured JSON logs with request ids and tenant ids.
• Service-level golden signals: latency, error rate, saturation, traffic.
• Dashboards for ingestion throughput and reconciliation drift.

Rollout checklist

• Health probes tuned to application cold start.
• Horizontal autoscaling rules (CPU + queue depth).
• Budget alerts; per-service cost attribution tags.
• Disaster recovery drills for config and data restores.

Results

Reduced Mean Time To Repair (MTTR), faster deploys, and a clear path to scale without re-architecting.