0

I want to incorporate A/B testing in courier service. So while sending goods there are api hits by different routing points, which are independent hit. Suppose A hits api1 which sit in M1 machine (old code) for sending a mobile. B hits api2 in M2 machine(new code) for receiving . New code in M2 expects a new parameter which old code in M1 doesnt send. There is a false failure here. What can be done?

Arko
  • 1
  • 1

1 Answers1

0

None of the existing tools will support a headless programmatic experiment, except Variant. (Disclaimer: I work there.) The reason is that all of them are SaaS and therefore, are instrumented at head of the stack in the UI tier. Variant server runs on your LAN and your host app can talk to it as a client. In this case, it sounds like you wold have the incoming api request come to one endpoint, where your experiment will be instrumented and forward on to either of the two services.

Igor Urisman
  • 717
  • 1
  • 6
  • 22