Building hybrid multi-tenant architecture for stateful services on AWS

This article describes a hybrid multi-tenant architecture for stateful services on AWS, specifically designed for large-scale ad-serving infrastructure.

• Replaced inefficient cellular architecture (one AWS account per tenant) with shared infrastructure model
• Reduced tenant onboarding time from 52 days to 7 days (86% reduction)
• Implemented three-level hierarchy: tiers, cells, and infra groups for scalable multi-tenancy
• Used Route 53 weighted routing to distribute traffic across multiple load balancers transparently
• Deployed dedicated ECS clusters per tenant within shared accounts for isolation
• Pre-wired AWS PrivateLink connections at tier creation, reducing setup steps by 80%
• Achieved up to 100 tenants per AWS account with strong cluster-level isolation
• Improved CPU utilization from 3% to significantly higher through efficient resource sharing

The architecture solves multi-tenancy challenges by decoupling dependency setup from tenant onboarding, enabling configuration-driven scaling without per-tenant infrastructure provisioning.