Demystifing the Hype around Modern Cluster Managers
summary
Containers become a prerequisite for cluster managers because they provide isolation and packaging that lets schedulers place workloads predictably, and microservices drive a rapid rise in container counts. Mesos exposes primitives like launch kill reserve and status update and relies on two-level scheduling so frameworks such as Marathon or Aurora implement very different DSLs. Roadmap items are hierarchical roles multi-tenancy revocable and scarce resources and new allocation algorithms, all aiming for higher utilization and lower hardware and energy waste.
AI-written summary · about
chapters (7)
- 0:00Start
- 2:14Background Information
- 5:47Examples from Real World
- 11:15Microservice
- 16:27Run Different Types of Workloads in the Same Cluster To Allow for Efficiency
- 27:04Unified in Stable Api'S
- 28:11Architecture of Our Data Centers
related talks
23mYou Ever Wonder Why We're Here? - Sarah Christoff, Cloudreach (Beginner Skill Level)
21mDDD and FSM: tackling complexity with state machines
22mRight Abstraction
38mIt's a Trap! How Abstractions Have Failed Us.
31mKubernetes on Top of Mesos on Top of DCOS
40mPanel Discussion: Multi-Tenancy in Kubernetes: Current State and Future Roadmap
35mInto the Shopfloor: Moving Manufacturing Execution Systems To Kubernetes
19mMultigrain Services