Tectonic Migrates Their Multi-Cloud Azure + AWS Setup to GCP with Facets

How Tectonic worked with Facets to bring its production workloads from AWS and Azure onto a single platform on Google Cloud, eliminating cross-cloud overhead, cutting event processing costs by 60%, and unlocking the headroom to grow customers without growing operations.

About Tectonic

Tectonic Technologies is a growth platform powering AI-driven customer experience and storefront solutions for high-growth Shopify-based merchants across Beauty and Personal Care, Fashion, and CPG. At the center of the platform is Hermes, a mission-critical event processing engine that ingests clickstreams, transactions, and behavioral data in real time to fuel personalization, intelligent recommendations, and marketing attribution.

The Challenge: A Fragmented Cloud Footprint

Tectonic was running production across two clouds. The core backend lived on AWS. Hermes, the event processing engine, ran on Azure. The setup had been functional, but as customer volume grew, the costs and complexity of operating across two clouds started compounding fast.

Three structural problems were eating into margins and slowing the business:

• Expensive hosted Azure Functions. Hermes processed huge volumes of events, and the per-execution consumption model became prohibitively expensive at scale.
• Cross-cloud egress costs. Moving event data between Azure and the AWS-hosted backend generated significant recurring overhead with no business value attached to it.
• Inefficient multi-tenancy. Every customer (large or small) had a dedicated queue and a dedicated set of consumers, leaving smaller tenants severely under-utilizing expensive infrastructure.

On top of that, the AWS backend had grown organically over the years. Infrastructure changes required senior engineers. Every new customer onboarding involved bespoke infrastructure work. And the per-customer cloud isolation model demanded a reproducible architecture, not a one-time lift.

The strategic question was no longer whether to consolidate. It was how fast Tectonic could get there without disrupting the business.

The Solution: A Single Platform on Google Cloud

Tectonic engaged Facets to consolidate everything onto Google Cloud, and to do it without freezing the product roadmap. Facets ran the work across two parallel tracks.

Track 1: Hermes Event Processing, Azure to GCP in 2 Weeks

Hermes was the highest-leverage cost problem. Facets re-architected it on a modern Google Cloud foundation:

• Managed event streaming replaced Azure Functions and bespoke queue management. Fully managed, infinitely scalable, no operational overhead.
• Containerized compute on Google Kubernetes Engine drove compute costs down through fine-grained resource control and intelligent use of spot capacity.
• Backlog-driven autoscaling scales consumer workloads up and down based on live event volume, including scale-to-zero during idle windows.
• Optimized multi-tenancy. Smaller tenants now share queues and consumers, dramatically improving resource utilization. High-volume tenants retain dedicated infrastructure for performance isolation.

The migration completed in two weeks, with no customer impact. The cost of event processing infrastructure dropped 60%.

Track 2: Full Backend Platform, AWS to GCP in 90 Days

In parallel, Facets ran the broader consolidation of Tectonic's main backend onto Google Cloud, using a declarative blueprint of the target architecture:

• Every service, database, and dependency now lives in one Facets blueprint that powers every customer environment, every region, and every release.
• All production databases were moved in a single 50-minute cutover, with a tested rollback held in reserve throughout.
• The same blueprint provisions production (US), production (India), and development consistently, with policy and cost controls applied across all of them.

Throughout both migrations, Tectonic engineers were deploying releases independently. Work that previously required senior-engineer involvement became self-service from day one.

What Made the Difference

This wasn't a tooling change. It was an operating model change.

The technical migration mattered, but the strategic outcome was that Tectonic stopped operating multiple clouds with multiple operating models and started operating one platform with one operating model. The same blueprint that ran the migration runs new customer environments. The same governance that controls production releases controls every new tenant onboarding.

For Tectonic, the result is a platform that compounds over time:

• New customers come online without bespoke infrastructure work
• Architectural changes roll out across all environments from a single blueprint
• Cloud spend stays aligned to actual usage, not worst-case provisioning
• Engineering velocity isn't gated by a small group of senior operators

"The 60% cost reduction was the headline. The bigger unlock was operational. Releases are self-service, new customers are onboarded without infrastructure work, and we're spending engineering time on the product roadmap instead of infrastructure rebuilds." - Nishant Khurana, Co-founder & CTO, Tectonic

"

Strategic Value Delivered by Facets