Skip to content

Latest commit

 

History

History
49 lines (37 loc) · 2.75 KB

README.md

File metadata and controls

49 lines (37 loc) · 2.75 KB
Raw

...a reverse proxy that acts as an intermediary between your apps and the API of your Feature Flag SaaS Tool, reducing costs by centralizing feature flag evaluations and avoiding vendor lock-in.

Feature Flags Proxy

  • Feature Flag Evaluation: Exposes an endpoint to check the status of a feature flag.

    • Example: curl "http://localhost:8080/feature/feat1?microserviceId=curlcommand"
    • Response: Returns True or False based on the feature flag status for the feature nammed feat1 on your feature flag tool (unleash, launchdarkly etc).

  • Vendor Lock-in Avoidance: This project abstracts the feature flag evaluation logic through a proxy, making it easier to switch between feature flag tools (e.g., from Unleash to another service) without making significant changes to your underlying codebase. This helps prevent vendor lock-in and provides flexibility in choosing or changing feature flag providers.

  • Caching: Implements a caching layer for feature flag results with a default Time-to-Live (TTL) of 2 minutes, configurable via environment variables.

  • Logging: Logs request details including:

    • The IP address of the requestor.
    • Custom source information (e.g., label myapp-prod) passed through the request for traceability.

  • Context Support (In Progress): Planned support for passing user-specific data (contexts) to the underlying feature flag tool. This will be crucial for enabling A/B testing, as it allows the flag status to differ based on the user or client context.

  • Stack: Built in Java and Dockerized, self-hostable

Commands to run and test the app locally

docker build -t feature-flags-proxy .
docker run --rm -d -p 8080:8080 \
  -e LAUNCHDARKLY_SDK_KEY="sdk-XXXX" \
  -e CACHE_TTL_MINUTES=1 \
  --name feature-flags-proxy-container \
  feature-flags-proxy
curl "http://localhost:8080/feature/feat1?microserviceId=curlcommand"
curl -f http://localhost:8080/actuator/health

TODO

Can be considered as production ready, but here is a todo of improvements:

  • Implement multi-context support for A/B testing (allow different flag states based on user data). The best way to implement this seems to be by passing a JSON payload to the reverse proxy that contains the relevant data.
  • Implement the SDK of your given feature flagging SaaS tool!
  • Implement CI/CD pipelines for Docker image building and deployment.
  • Optimize caching mechanism? (event-driven cache invalidation, distributed caching, fine-grained caching strategy, LRU?, LFU?)
  • Improve logging and tracing.s
  • High availabilty
  • IP rotation - proxy pooling
  • Throttle requests
  • Randomizing headers and user contexts for each request
  • Improve security posture (docker, TLS etc)

Screenshot