Project Name
Fintech Scaled to 200+ Microservices Without Losing Control Using Backstage and Dagger
A fast-growing fintech with 200+ microservices on OCI Kubernetes Engine had scaled rapidly with no central service registry, no shared CI/CD standards, and no ownership mapping. Engineers spent hours tracking down service owners, new hires took weeks to become productive, and every team maintained its own bespoke YAML pipeline that only the original author understood. Deployment drift made every incident response unpredictable. Applying its AI-First approach, Ksolves deployed a two-pillar platform engineering strategy: Backstage as the unified developer portal and service catalog, and Dagger as the standardised, portable CI/CD engine.
- Zero Service Discoverability: No central registry existed to map 200+ microservices to their owners, dependencies, documentation, or operational status. Engineers routinely spent hours tracking down the right team for a single question.
- Weeks-Long Onboarding Cycle: New engineers faced 3-4 weeks of onboarding because tribal knowledge was the only way to understand which services existed, how they interconnected, and where their documentation lived.
- Bespoke YAML Pipeline Per Team: Each team maintained its own hand-crafted CI/CD pipeline, dozens of unique, untestable, and drift-prone YAML definitions with no shared standards, no reuse, and no way to test without pushing to CI.
- Deployment Drift Across Teams: Without a standardised build-test-deploy pattern, identical microservices built by different teams produced different artifacts, used different testing gates, and deployed through different mechanisms, making incident response unpredictable.
- No Ownership or Dependency Mapping: Service-to-team ownership was undocumented. Dependency relationships between microservices were unknown until something broke in production, turning every incident into a discovery exercise.
- Local Pipeline Testing Was Impossible: YAML pipelines could only be tested by pushing to CI - every iteration required a git commit, extending developer feedback loops from minutes to hours and slowing every pipeline improvement cycle.
Ksolves deployed a two-pillar strategy: Backstage as the unified service catalog and Dagger (Python SDK) as the standardised CI/CD engine. One catalog, one pipeline pattern, one verified deployment path, regardless of which team built the service.
- Backstage Service Catalog on OKE: Deployed Backstage as the central developer portal integrated with OKE, mapping every microservice to its owning team, upstream/downstream dependencies, API documentation, and operational runbooks - eliminating the service discovery problem.
- Dagger (Python SDK) Pipeline Standardisation: Replaced all bespoke YAML pipelines with Dagger-based CI/CD definitions written in Python - making every microservice build locally testable, structurally identical, and portable across CI environments without rewriting.
- OCI DevOps and OCIR Integration: Configured OCI DevOps as the deployment pipeline orchestrator with OCI Container Registry as the single container image store - ensuring every team pushed artifacts through the same verified promotion path.
- Backstage Scaffolder for Golden Paths: Implemented Scaffolder templates encoding production-ready service creation patterns - new microservices came pre-wired with ownership metadata, CI/CD pipeline, documentation stubs, and OKE deployment manifests from day one.
- TechDocs for Docs-as-Code: Integrated TechDocs into Backstage so documentation lived alongside source code in Git and was automatically published to the service catalog - eliminating stale wikis and scattered Confluence pages across all teams.
Technology Stack
| Category | Technology |
|---|---|
| Platform | Backstage |
| Processing | Dagger (Python SDK) |
| Infrastructure | OCI Kubernetes Engine (OKE) |
| Integration | OCI DevOps + OCIR |
| Methodology | Backstage Scaffolder + TechDocs |
- Service Discovery Cut to 10 Seconds: Backstage catalog returns ownership, dependencies, and docs in under 10 seconds - replacing 30-60 minutes of Slack threads and tribal knowledge per query.
- Onboarding Reduced by 70%: Self-service catalog, TechDocs, and golden-path scaffolding cut onboarding from 3-4 weeks to under one week - no tribal knowledge transfer required.
- 90% of Pipeline Definitions Consolidated: One Dagger-based pattern replaces 200+ bespoke YAML files - a single shared, versioned library with locally testable, identical builds across all teams.
- Zero Deployment Drift: Every service builds identically through Dagger and deploys through OCI DevOps to OCIR - identical artifacts, identical gates, predictable incident response.
- Feedback Loop From Hours to Minutes: Dagger pipelines run identically on developer laptops, pipeline changes are tested locally in minutes, no git push to CI required.
“For the first time, every engineer can find any service, understand who owns it, and deploy it the same way. Backstage and Dagger gave us a platform, not just pipelines.”
– VP of Engineering.
A fintech running 200+ undiscoverable microservices with fragmented YAML pipelines and weeks-long onboarding was transformed into a governed, developer-first platform through Ksolves’ DevOps consulting services. Backstage unified service discovery and ownership into one catalog. Dagger replaced 200+ bespoke pipelines with one locally testable build pattern. Onboarding dropped by 70%, pipeline definitions consolidated by 90%, and deployment drift was eliminated – with the foundation ready to extend into compliance gates, FinOps attribution, and self-service environment provisioning.
Scaling Fast but Losing Control of Your Microservices? Let’s Fix it.
