STRV dumps Fargate monolith for serverless - weekend hackathon validates migration path
Czech software firm STRV migrated an internal project from a Fargate-hosted monolith to AWS serverless architecture - but the how is more interesting than the what.
Engineering Manager Marek Čermák brought in AWS consultant Michal (via AWS Hero Filip Pýrek) not to deliver a finished architecture, but to run ongoing knowledge-transfer sessions. Over several months, STRV's team explored AppSync, Step Functions, DynamoDB, and Glue through real implementations, not slide decks.
The original stack - Fargate container, Aurora, S3 - works fine. STRV still uses it. But this specific project needed something leaner. The goal: lower operational overhead, better cost efficiency, faster iteration.
The validation approach matters here. Rather than commit to a full rewrite, the team ran a weekend hackathon in their Prague office. Two teams, same challenge, different approaches. One went pure serverless (AppSync + DynamoDB). The other blended Lambda with familiar tooling - incremental change over revolution.
Both prototypes worked. That's the point.
What this means in practice
The broader pattern aligns with what we're seeing across enterprise AWS migrations. Shutterfly's recent shift cut operating costs 25% and eliminated VMware licensing entirely - but only after moving 800 systems and 400TB of data. STRV's approach inverts that risk profile: validate small, learn fast, then scale.
Worth noting: AWS has been pushing "cloud enablement engines" - reusable migration patterns that accelerate subsequent moves. STRV essentially built their own through hands-on exploration. Principal Engineer Tomáš Kocman called it team-building as much as technical upskilling.
The real test comes next: whether this prototype graduates to production, and whether the patterns transfer to client work. The monolith-to-microservices path is littered with abandoned rewrites. STRV's bet is that developer confidence - earned through building, not briefings - changes those odds.
No word yet on production timelines or cost savings. We'll see if this stays an experiment or becomes their new default.