Harness cloud elasticity by bursting to Amazon EC2 Spot
summary
Wētā FX pushes final high‑fidelity renders to Amazon EC2 Spot, scaling from a 45–50M thread‑hour on‑prem baseline to peaks around 130 million thread hours per week and delivering over a billion VCPU hours across regions; outcomes: the film finished on schedule and four other shows ran concurrently. Key tools are Spot Blueprints and attribute‑based instance selection to diversify capacity, and operations tolerate Spot’s two‑minute interruption notice with fault‑tolerant workflows and storage tuned for long‑distance latency.
AI-written summary · about
chapters (6)
- 0:00Introduction and context
- 3:41Cloud burst concept explained
- 8:23EC2 Spot instances overview
- 12:02Wētā FX case study
- 24:37Tools for cloud bursting
- 27:32Summary and resources
related talks
42mAWS Spot instances: huge savings (or huge risks?)
47mReliable, secure, and efficient cloud operations
31mKeeping Your Cloud Footprint in Check
46mExtreme infrastructure automation with Wavefront
43mSpot Instances
10mLightning Talk: Does Cloud Elasticity Pay off for HPC Workloads?
34mHPC on AWS: Innovation without infrastructure constraints
32mBlending Spot and On-Demand Instances and Optimizing Costs