From UBER TECHNOLOGIES INC
Hongliang Xu is a software engineer with streaming data team at Uber, primarily focusing on developing and maintaining Uber’s s streaming data infrastructure built using Apache Kafka, which is a critical component of Uber’s data pipeline handling billions of messages and hundreds of terabytes of data per day.
Most storage systems replicate data across data center asynchronously, which introduces eventual consistency in the system. This is not acceptable for high-value data which requires strongly consistency across data center. At Uber, we build a messaging system that can synchronously replicate data to another data centers such that data will be available even after single data center failure. In this talk, we will discuss how we leverage Apache Zookeeper, Apache Kafka and cloud to build this system.