Case Study: Qunar achieves 300x faster real-time analytics with Alluxio

Qunar Performs Real-Time Data Analytics up to 300x Faster with Alluxio

Qunar, a leading travel search engine in China, needed to speed up its real-time analytics platform, which processes about 6 billion log entries a day. Its existing Spark Streaming and Flink architecture relied heavily on remote HDFS storage, creating network latency, disk I/O bottlenecks, checkpointing issues, and garbage collection problems that hurt stability and low-latency performance.

To solve this, Qunar deployed Alluxio as a storage management layer between its compute frameworks and HDFS, using Alluxio’s memory-speed, tiered storage, unified namespace, and HDFS-compatible API. With Alluxio, Qunar reported average job times dropping from 8 minutes to about 30 seconds, throughput rising from 20–300 events per second to around 7,800 events per second, and peak performance improvements of up to 300x.

Qunar