What Is the Cloud?
summary
Cloud here means taking standard rack servers (e.g., 48 CPUs, 384 GB RAM, ~12×3 TB) and slicing them with virtualization so VM images and block volumes are managed by OpenStack components (Glance, Cinder, Neutron, Nova). The talk covers the need for live‑migration and shared storage to survive host failures, the requirement for high bisection bandwidth in the network (Terasort exposes bottlenecks), and mentions networking/control technologies (Contrail BGP/MPLS, Midonet, Open vSwitch) plus consensus systems as source‑of‑truth (ZooKeeper/Consul). It also warns that missing quotas, bad defaults and untested distributed designs lead to real operational problems.
AI-written summary · about
chapters (12)
- 0:00Start
- 3:30We are virtualizing!
- 4:42What does an instance need to boot?
- 6:06What if a host goes down?
- 8:36Hardware as a script, Infrastructure as Code...
- 12:40I can haz live migration, plz?
- 18:08Capacity problem? What capacity problem?
- 18:44Terasort to watch the world burn
- 19:44Meanwhile, at the Chocolate Factory...
- 21:25Contrail
- 29:13Scheduler, Spread Strategy
- 35:24Ephemeral vs. Persistent Volume
related talks
33mWhat the Cloud?
21mBack to the original mission: an open cloud operating system
24mHow to survive Cloud – An Ops perspective
22mCloudy Operations
21mHybrid Cloud - nothing to worry about! 🌥️
18mThe Cloud Myths, Schemes, and Dirty Little Secrets
22mFrom Bare Metal to Abstraction: A Tale to the Heart of Cloud-Native Computing
26mCloud Agnostic Webscale