Mesos, Then and Now
summary
Apache Mesos begins as Berkeley’s Nexus in 2010 and becomes Twitter’s cluster fabric by 2012, paired with Aurora scheduler to run JVM services and improve packing across 64+ core hosts; early experiments use LXC and an internal bundler called app app, and Docker adoption is deliberately delayed so teams ship jars on host filesystems. Mesosphere commercializes Mesos; Twitter runs one Aurora instance per cluster, pushes for stateful frameworks, fixes scale bugs (master failover TCP floods) and pursues better resource isolation to raise utilization.
AI-written summary · about
chapters (13)
- 0:00Introduction and setup
- 1:15Apache Mesos creators introduction
- 3:19Early Mesos adoption at Twitter
- 8:46Containers and Docker at Twitter
- 12:38Why containers and Mesos
- 19:01Aurora and Mesosphere's role
- 22:35Mesos architecture explained
- 27:17Two-level scheduling innovation
- 30:05Who runs Mesos today
- 34:58Resource utilization vs pipeline enablement
- 38:30Twitter’s scale and innovation
- 40:35Future of Mesos at Twitter
- 43:51Collaboration and Q&A
related talks
59mBuilding Distributed Systems with Apache Mesos
42mMesos: The Operating System for your Cluster
57mA Distributed and Fault-Tolerant Application with Apache Mesos & Apache Aurora
53mDocker as the Building Block for Datacenter-Scale Applications
01h 00mDeveloping Distributed Systems with Apache Mesos
22mKeynote: The Future of Apache Mesos and DC/OS
23mHow cluster managers affect the landscape of modern distributed computing
13mKeynote: Welcome and State of Mesos
