Explore the potential of Docker for datacenter-scale applications in this 53-minute conference talk from GOTO Amsterdam 2015. Delve into the challenges of managing large-scale container-driven clusters and learn how leading companies automate their operations. Discover how to build distributed systems using Docker, and understand the role of Apache Mesos in creating highly-elastic, highly-available systems. Examine topics such as containerization, failure management, cluster management architecture, resource scheduling, and datacenter operating systems. Gain insights into Mesos primitives, task isolation, failover strategies, resource reservations, and persistent volumes. Learn how to leverage these concepts to create efficient, scalable applications across entire datacenters or cloud environments.
Docker as the Building Block for Datacenter-Scale Applications
Overview
Syllabus
Introduction
an emerging trend
containerization
a reinforcing trend
failures
maintenance upgrading software i.e., the kernel
challenges
planning for failure
different software
Mesos is a cluster manager with a master/slave architecture
schedulers register with the Mesos masters in order to run jobs/tasks
service scheduler
configuration/package management
multiple schedulers
Mesos: datacenter kernel
Mesos primitives
a task with a command
a task with an executor
task/executor isolation
master failover
scheduler failover
slave failover
resource reservations
static reservations
reservations provide guarantees, but at the cost of utilization
inverse offer
persistent volumes
need a datacenter operating system
DCOS CLI
Taught by
GOTO Conferences
