Rise of the Machines [B]
summary
Workable, concrete takeaway: a tiny custom scheduler (a few hundred lines of Go) ties Kubernetes to SQS metrics and a two-tier microservice (quick vs slow work) and runs containers every few seconds to meet SLAs while maximizing bin‑packing. The live micro-batcher proof‑of‑concept plugs into production and reports a 70% cost reduction in queue-backed processing by stopping and starting low‑priority jobs dynamically. Covers how to build it, what APIs are used, and why this automation reshapes ops.
AI-written summary · about
chapters (14)
- 0:00Introduction
- 0:25Agenda
- 7:00Humans anthropomorphize
- 12:45Microscaling
- 17:35MicroBatcher
- 18:12Micro Services
- 22:47Custom Scheduler
- 24:40Automation
- 27:00Kubernetes
- 27:31Is Kubernetes evil
- 27:43The future of data centers
- 28:18Three areas of automation
- 29:01Is automation always good
- 29:57What should not be automated
related talks
24mKubernetes Batch Processing at Scale - A Scheduling Perspective
37mRunning Kubernetes in a Manufacturing Line
53mMaking tomorrow's code look like today's
38mCost-Effective Scheduling of a Massive Number of Containers in Kubernetes
33mFrom spreadsheet scheduling to Kubernetes: building an on-premise ML platform
47mRise of the Machines
59mHow Kubernetes optimisation can combat climate change
51mCode the Future, Now