Making the LinkedIn experimentation engine 20x faster
At LinkedIn, we like to say that experimentation is in our blood because no production release at the company happens without experimentation; by “experimentation,” we typically mean “A/B testing.” The company relies on employees to make decisions by analyzing data. Experimentation is a data-driven foundation of the decision-making process, which helps with measuring the precise impact of every change and release, and evaluating whether expectations meet reality.
LinkedIn’s experimentation platform operates at an extremely large scale: It serves up to 800,000 QPS of network calls, It serves about 35,000 concurrently running A/B experiments, It handles up to 23 trillion experiment evaluations per day, Average latency of experiment evaluation is 700 ns and the 99th percentile is 3 μs, It is used in about 500 production services. It is used in about 500 production services.