Completed
How do we mitigate risks with 1M machines per deployment?
Class Central Classrooms beta
YouTube videos curated by Class Central.
Classroom Contents
Twine - A Unified Cluster Management System for Shared Infrastructure
Automatically move to the next video in the Classroom when playback concludes
- 1 Intro
- 2 Data center geographic regions
- 3 What design decisions did Twine make differently?
- 4 What if we used Kubernetes?
- 5 How does Twine avoid stranded capacity?
- 6 How does Twine perform fleet-wide optimization?
- 7 How does Twine perform fleet-wide optimization fo. entire geographic region?
- 8 How well does the Twine scheduler scale?
- 9 How do we mitigate risks with 1M machines per deployment?
- 10 Private pools or shared infrastructure?
- 11 What is host customization?
- 12 What is the overhead for host profile switches?
- 13 What drives host profile changes?
- 14 What are the challenges with supporting ubiquitous shared infrastructure?
- 15 Challenge: Tasks are not homogenous
- 16 How does Twine collaborate with applications?
- 17 What is our shared infrastructure adoption?
- 18 How easy is it to migrate onto shared infrastructure.
- 19 Power is our most constrained resource
- 20 Big machines or small machines?
- 21 Why use small machines?
- 22 How much do we save by using small machines?
- 23 What lessons did we learn using small machines?
- 24 Conclusion
