REST API Performance Testing
Across European Cloud Providers

Project Description

The goal of this project is to evaluate how different REST API design patterns (REST, GraphQL-over-HTTP, chunked streaming responses) affect client-perceived latency under realistic network conditions. We deliberately test both short-lived request/response cycles and long-polling / streaming endpoints to understand how servers behave under sustained connections.

Our test runner sends scheduled HTTPS requests every few seconds from our Frankfurt node to a list of target APIs. Some tests use persistent connections with keep-alive, others deliberately close and re-open TCP sessions to measure handshake overhead. All traffic is HTTPS on port 443.

Tested Endpoint Types

• GET
  /api/v1/status

  Health check - short response, measures baseline RTT
• POST
  /api/v1/echo

  Echo payload - tests request/response body serialization time
• GET
  /api/v1/stream

  Chunked streaming - holds connection open, measures TTFB vs. completion
• POST
  /api/v1/batch

  Batch processing - large payload, tests throughput limits
• GET
  /api/v1/metrics

  Prometheus-style metrics scrape - periodic, every 30s
• PUT
  /api/v1/session

  Session keepalive - long-lived connection test, up to 5 min