Docker as the Building Block for Datacenter-Scale Applications
summary
Explains how Apache Mesos becomes a datacenter kernel that runs Docker or tarball workloads with two-level scheduling, offering concrete tools: Marathon and Kubernetes run as schedulers, DCOS CLI packages services, and Mesos primitives include persistent volumes, revocable resources and dynamic reservations. Describes Twitter’s move from a Rails monolith to microservices on Mesos, Apple running Siri on Mesos, and features that work in production: master/scheduler/slave failover, vertical scaling on executors, inverse-offers for maintenance and guaranteed resource reservations.
AI-written summary · about
chapters (29)
- 0:00Introduction
- 0:30an emerging trend
- 2:43containerization
- 4:12a reinforcing trend
- 6:04failures
- 6:54maintenance upgrading software (i.e., the kernel)
- 9:13challenges
- 9:28planning for failure
- 11:31different software
- 14:10Mesos is a cluster manager with a master/slave architecture
- 14:40schedulers register with the Mesos master(s) in order to run jobs/tasks
- 15:16service scheduler
- 16:46configuration/package management
- 19:35multiple schedulers
- 22:37Mesos: datacenter kernel
- 23:49Mesos primitives
- 29:26a task with a command
- 30:05a task with an executor
- 31:14task/executor isolation
- 33:00master failover
- 33:33scheduler failover
- 35:05slave failover
- 37:28resource reservations
- 38:03static reservations
- 39:00reservations provide guarantees, but at the cost of utilization
- 42:03inverse offer
- 43:13persistent volumes
- 45:34need a datacenter operating system
- 46:32DCOS CLI
related talks
39mHow we learned to love the Data Center Operating System
12mEdward Hsu, Mesosphere | DockerCon 2018
40mFrom Monolith to Docker Distributed Applications
22mKeynote: The Future of Apache Mesos and DC/OS
59mBuilding Distributed Systems with Apache Mesos
42mMesos: The Operating System for your Cluster
01h 00mDeveloping Distributed Systems with Apache Mesos
48mContainer: check! Aber wohin mit Persistent Data, Big Data oder Fast Data?!
