vBOIDs: Taming Chaos via Coarse-Grained Scheduling Abstraction for Containers
Talkat 20th USENIX Symposium on Operating Systems Design and Implementation (OSDI '26) · 2026-07-13..07-15·
summary
Introduces a kernel patch (≈2,000 LOC on Linux 6.15) that groups container threads into lightweight placement units called BOID abstraction and coordinates a global and local policy via two-level balancing. Implements a two-phase move (atomic target update then deferred per-thread migration) exposed through a cgroup CPU file to create N BOIDs. Shows concrete wins: cross-core migrations drop from ~316 to ~81/s/core in an example, sustains 4,000 rps, hits up to 3× throughput and keeps P95 latency <500 ms on Google Boutique.
AI-written summary · about
chapters (10)
- 0:00vBOIDs introduction
- 1:24Containers scheduling chaos
- 3:54VMs vs containers comparison
- 6:25vBOIDs approach
- 8:56vBOIDs implementation
- 10:21Evaluation setup
- 11:25Hotel workload results
- 12:26Social network results
- 12:46Online boutique results
- 13:29Summary
related talks
44mUserspace Assisted Scheduling via Sc...
05mCombat Chaos-Driven Delivery By Thinking Like An OS
39mPractical Container Scheduling: Juggling Optimizations, Guarantees, and Trade-Offs at Netflix
35mLeveraging the Kubernetes Enqueue Scheduling Plugin for Smarter Workload Placement
29mTowards Container layer aware Scheduling Policies
39mOptimizing CephFS with Combining MDS QoS Scheduling and Static-Dynamic Subtree Partitioning
46mZbus - New Features and Roadmap
42mInter Process Confusion - Navigating Zephyr's IPC APIs