Kingsum has been a principal engineer at Alibaba Cloud Intelligence System Software Hardware Co-Optimization driving performance optimization at the scale of data centers since 2016. He also represents Alibaba in the Java Community Process Executive Committee since 2018. Since receiving Ph.D. in Computer Science and Engineering from the University of Washington in 1996, he has been working on performance, modeling and analysis of software applications. He has been issued more than 20 patents. He has presented more than 100 technical papers. He appeared four times in JavaOne keynotes. In his spare time, he volunteers to coach multiple robotics teams to bring the joy of learning Science, Technology, Engineering and Mathematics to the K-12 students in USA and China.
Java is often criticized for hogging memory resources. eCommerce applications may need to process very different amount of traffic between day and night. As we have a very large number of Java applications running in the biggest global shopping festival, optimizing the use of memory in Java heaps would lead to significant cost saving.
In this presentation, we share our development of our ElasticHeap memory management technology. The approach we took can adjust the heap usage based on the GC pressure without adversely affecting stop-the-world time like other scaling methods. In addition, we also return memory to the system when there is little traffic. We achieved more than 80% memory saving among thousands of Java application instances in our 11.11 global shopping festival.