Apache projects, let's embrace ARM!
summary
Effort adds real ARM CI and binaries across Apache: Hadoop ships Hadoop 3.3.0 ARM binaries after fixing dependency breaks (protobuf upgrade, node/native libs), replacing embedded test libraries with Testcontainers or disabling tests when no arm image exists, and donating CI servers plus OpenLab access. CI runs on Jenkins agents, Travis CI and GitHub Actions via a remote Kubernetes bridge at Huawei and self-hosted nodes (Jenkins/Travis/GitHub Actions via remote k8s). Benchmarks show TeleSort end‑to‑end ~40% faster on ARM and library tweaks (zstd, snappy) give ~10–13%–plus gains.
AI-written summary · about
chapters (8)
- 0:00Start
- 0:47Speakers Introduction
- 5:18ARM Chips
- 10:01Case1: Hadoop ARM support
- 13:30Case 1 Hadoop ARM Support: Compression Performance Tuning
- 14:47Case1: Hadoop Compression performance Tuning
- 26:40ARM64 HARDWARE FOR CI
- 27:01Problems and obstacles
related talks
26mMake life easier for Big Data users on ARM - Our efforts and future plans
52mUbuntu on ARM
41mRunning Java on ARM - is it worth it?
42mJava on ARM. Theory, Applications and Workloads
39mJava on ARM. Theory, Applications, and Workloads
31mContinuous Delivery on Arm: A Practical Approach
27mWhat's Going ARM: Adopting ARM64 At Airbnb
43mNative CI and CD pipelines for Arm