OrderServ's transition from dedicated ECS clusters per brand to a multi-tenant shared cluster model has significantly lowered operational expenses, improved scaling responsiveness, and streamlined developer workflows. This architecture evolution supports rapid onboarding of new brands while maintaining strict data isolation and high availability during traffic surges.
- 64% reduction in compute costs and 80% lower authentication expenses
- Scaling times decreased from 5–7 minutes to 30–60 seconds during demand spikes
- Shared ECS cluster with tenancy library enables configuration-driven onboarding
Infrastructure signal
OrderServ's infrastructure shifted from isolated ECS clusters for each brand to a consolidated multi-tenant cluster, reducing the duplicated resource overhead associated with running 21 microservices per tenant. This shared cluster model enables economies of scale by hosting multiple brands together while still allowing high-volume tenants to opt out for dedicated clusters when necessary.
The platform maintains strict data isolation through per-tenant data stores using Amazon DynamoDB, Amazon Aurora, and ElastiCache (Valkey), supported by AWS Key Management Service encryption and Transport Layer Security. Isolation is primarily enforced via a shared .NET Core tenancy library that manages tenant resolution and resource routing, with IAM and resource policies serving as safety nets.
Developer impact
Developers benefit from a streamlined deployment and configuration approach as new brands can be onboarded largely through configuration changes instead of provisioning separate infrastructure. This reduces time-to-market and operational complexity, allowing teams to focus on feature development rather than cluster management.
The tenancy library centralizes multi-tenant logic, simplifying coding standards around tenant context propagation and resource access. This approach enables faster iteration cycles, reduces potential multi-tenancy bugs, and ensures consistent enforcement of security and data access policies across all microservices.
What teams should watch
Operations and platform teams must monitor scaling performance closely, as the new architecture has dramatically reduced scaling latency from several minutes to under a minute during traffic surges, which is critical for handling unpredictable event-driven demand spikes typical in quick-service restaurant environments.
Security teams should remain vigilant about the tenancy library's integrity and the effectiveness of IAM policies, given that infrastructure isolation is loosened in favor of library-enforced tenant segmentation. Robust observability into authentication and service-to-service traffic using encrypted communications will be essential to detect and mitigate risks.
Databases and backend platform teams need to ensure that per-tenant data stores maintain performance and strict isolation as brand count grows. Continued evaluation of shared versus dedicated cluster eligibility criteria will be important to balance cost savings with reliability and data security as the platform scales.