Rising Star Talk 2: Wednesday (August 24, 2022) 2:30 pm - 3:00 pm

Abstract:

There have been a lot of research efforts in improving the performance of graph processing. Nearly every work more or less makes some implicit assumptions that appear quite reasonable under research settings, and many proposed techniques depend on some subtle implications behind these assumptions. Unfortunately, many of them are at least partially invalid from industrial views. In this talk, I will introduce the story of how we adapt Gemini [OSDI 2016], an academic prototype graph processing system, to industrial application at Ant Group, showcasing the issues brought by some failed assumptions and our solutions to these new challenges. I will also show the other side of the coin, i.e. some opportunities to change industrial practice with the help of innovative research ideas.