Rich-text Reply

Java SDK server side how does the Experiment Traffic Allocation works in multi clusters

anacharya 01-05-17

Java SDK server side how does the Experiment Traffic Allocation works in multi clusters

Looking at the Java SDK uses MurmurHash3 function for user_id.

 

However, I am not able to understand once the hash is obtain how it determine the bucket in multi nodes environement if I have configured 50% 50% for two variations in the configuration of optimizely.

tedroddy 01-06-17
 

Re: Java SDK server side how does the Experiment Traffic Allocation works in multi clusters

We have something called deterministic bucketing that computes a variation based on the activated user_id and the traffic allocation configuration contained in the datafile. You can activate an experiment on any node, and if the user_id and datafile are the same, you'll always get the same variation.
Best,
Ted
Optimizely