From Zero to Infinity: How AI-Powered Hedge Fund Build Cloud-Native AI Platfo...
Talkat KubeCon + CloudNativeCon + Open Source Summit China 2023 · on Wednesday, 2023-09-27· at 14:45· 2层 会议室 2 | 2F Room 2·
summary
Describes how Metabit Trading built an elastic distributed cache on Kubernetes using Fluid (CNCF) and JuiceFS to solve 0→hundreds-of-pods scale‑up bottlenecks when many pods concurrently read from slow, rate‑limited distributed storage. Explains using Prometheus-driven cache‑usage behavioral metrics to autoscale the cache cluster to sustain up to 1000 Gbps, and using Fluid with CronHPA for scheduled scaling to trade off cost and performance, with performance and cost‑benefit evaluation and a demo.
AI-written summary · about
related talks
05mProject Lightning Talk: Fluid Data Anyway, Data Anywhere, Data Anytime
35mFluid: Kubernetes Native Distributed Dataset Orchestrator and Accelerator
31mGotta Cache 'em All: Scaling AI Workloads With Model Caching in a Hybrid...
26mSpeed up Your ML Workloads With Kubernetes Powered In-memory Data...
36mAn almost perfect match: cloud native and scientific computing
32mDesign Patterns for Running AIML and Bigdata Workloads on Kubernetes
40mOptimizing AI/ML Workloads
27mFrom Supercomputing to Serving: A Case Study Delivering Cloud Native Foundation Models