8 Million Tasks Daily: Driving Efficiency in Multi-Tenant Data Platforms at Ho...
Talkat KubeCon + CloudNativeCon + OpenInfra Summit + PyTorch Conference China 2026 · 2026-09-07..09-09·
summary
Horizon Robotics reworked its data platform to run 8 million short-lived tasks per day by replacing per-workflow agents with a centralized Argo Workflows Agent Service, cutting infrastructure and CPU use. They also refactored metadata and reduced VolcanoJob size to fix an Informer-induced bottleneck (bringing sync latency from 4 minutes to 10 seconds) and used Kueue plus Volcano to implement elastic quotas and handle preemption for multi-tenant stability.
AI-written summary · about
related talks
27mEmpowering Autonomy: BYD's Journey Taming Million-Task Scale With Argo...
29mLarge Scale Batch Processing with Argo Work...
25mOne Engine To Rule Them All: Unifying Cloud Workloads With Argo Workflows
29mBuilding a Batch Processing Platform...
27mBuilding a Data Science Platform with Argo Workflows
20mDemystifying Argo Workflows: An Architectural Deep Dive
08mLightning Talk: Enabling Hybrid Cloud Infrastructure with Argo Workflows
30mFrom API Events To Production in Minutes: Backstage as the Front Door...
