Make life easier for Big Data users on ARM - Our efforts and future plans
summary
Team adds ARM CI enabled for Hadoop Spark HBase Flink Storm Kudu Impala, donates Jenkins and OpenLab ARM servers and pushes patches upstream to fix build and runtime failures so projects actually compile and run on ARM. They run 3-node Hadoop clusters with 20/50/100 GB MapReduce benchmarks and report ARM performance comparable and sometimes better than x86, plus micro-optimizations in core libraries that yield up to ~20% speedups and plans to publish ARM-friendly releases.
AI-written summary · about
chapters (9)
- 0:00Introduction and team overview
- 3:43Team members and roles
- 4:57Challenges for ARM support
- 7:07Infrastructure and CI contributions
- 9:39Big Data dependency fixes
- 13:35Hadoop performance testing
- 15:47Cloud and library optimizations
- 18:25Current work overview
- 21:58Future plans and Q&A
related talks
30mApache projects, let's embrace ARM!
52mUbuntu on ARM
29mThe Progress for Cloud Computing on Arm architecture
42mJava on ARM. Theory, Applications and Workloads
39mJava on ARM. Theory, Applications, and Workloads
16mChange Your Chip Architecture and Save the World
30mHome Lab to the Moon and Back
34mARM Migration Made Practical