This directory contains an experimental resource adapter for running complex experiments with Semian.

The ExperimentalResource class simulates a distributed service with multiple endpoints, each with configurable latencies following statistical distributions. This allows for testing various failure scenarios and performance characteristics.

1. Multiple Endpoints: Configure any number of endpoints, each with its own fixed latency
2. Statistical Distributions: Latencies are assigned based on statistical distributions
   • Currently supports: Log-normal distribution
3. Latency Bounds: Set minimum and maximum latency constraints
4. Fixed Latencies: Each endpoint gets a fixed latency at initialization for consistent behavior
5. Request Timeouts: Configure a maximum timeout for requests
   • Requests that would exceed the timeout sleep for the timeout period then raise an exception
   • Useful for simulating real-world timeout behavior
6. Baseline Error Rate: Configure a probability of request failure
   • Requests fail randomly based on the configured error rate
   • Failed requests throw RequestError exceptions after partial processing
7. Service-Wide Degradation: Degrade the entire service with optional ramp-up time
   • Latency degradation: Add fixed latency to all requests
   • Error rate changes: Modify error rate for the entire service
   • Gradual ramp-up: Both degradations support gradual transitions over time

See example_with_circuit_breaker.rb for usage:

bundle install
bundle exec ruby example_with_circuit_breaker.rb

Output: