The world of Linux Containers might be the hottest technology helping businesses to build cloud ready applications and services. In this talk we will provide the current status of OpenStack support for containers and containers support for OpenStack.In the first part we will dive into two container-focused OpenStack projects: Magnum and Kolla and show how Magnum as an API service can help to provide "Containers as a Service" on top of OpenStack leveraging the capabilities of Kubernetes, Docker, Heat and Flannel, and show how Kolla is going to improve OpenStack operations by containerizing OpenStack components into micro services for simplified upgrades and deployment consistency, portability and scaling.In the second part we will see Magnum in action and learn how to get the whole thing running on top of his own, on an existing OpenStack Kilo environment!
Speaker: Daniel Bäurer
Noch mehr Vorträge von uns: https://www.inovex.de/de/content-pool/vortraege/
1. Let’s talk about the Cloud
Part II : OpenStack and Containers
inovex Meetup Cologne
2015, May 27th
Arash Kaffamanesh (@kaffamanesh)
www.cloudssky.com
2. About OpenStack-X Meetup Group
● Founded: April 2014
● Meetups: 3
● Stackers: 118
● Location: Cologne
● Sponsors:
○ RDO Community
○ PLUMgrid
○ Clouds Sky GmbH
○ Still some more places are free ;-)
● Co-Organizers Wanted
3. Let’s talk about ...
● OpenStack and Containers
o About OpenStack
o A Container ...
o About Magnum (Demo)
o About Kolla
o OpenStack needs Containers
o Containers need OpenStack (?)
o Discussion : Do they need each other?
4. About OpenStack
● Open Source Cloud Operating System
● Was initiated by NASA and Rackspace in
June 2010
● Is the largest open source project helping to
build private cloud environment or to build
public cloud offerings
● OpenStack Foundation founded in 2012
● Over 1 million lines of code
11. Container History
● 1988: FreeBSD Jail
● 2005: Solaris Zones
● 2005: OpenVZ
● 2008: LXC
● 2013: Docker
● 2014: Rocket
● 2015: Wocker :-)
12. A Container ...
● Uses Kernel Namespaces
● Uses Linux Cgroups
● Uses Container Image (e.g. Docker Image)
● Has a Lifecycle
● Is a slice of the Kernel
● Is inexpensive and fast
● Could be unsecure
o docker daemon runs as root (libvirt
libcontainer)
o RKT is daemon less (systemd nspawn)
13. A Container ...
● Uses Kernel Namespaces
● Uses Linux Cgroups
● Uses Container Image (e.g. Docker Image)
● Has a Lifecycle
● Is a slice of the Kernel
● Is inexpensive and fast
● Could be unsecure
14. About Magnum
● Magnum is an API Service on OpenStack
● Provides CaaS (Containers as a Service)
● Integrates
o Heat (Orchestration)
o Kubernetes (Cluster Manager, Scheduler)
o Docker (Container Manager)
o Flannel (Overlay Network)
o Keystone and Neutron
● Uses
o Micro OSes (Atomic Host, CoreOS, ...) for
running Micro Services by Micro Teams :-)
17. Magnum Architecture
● There are 7 type of objects in magnum
○ Bay: collection of nodes
○ BayModel: template for bays (define coe)
○ Node: VM or bare-metal running MicroOS
○ Pod: collection of containers
○ Service: abstraction of pods + policy to
access pods
○ ReplicationController: manages pods and
makes sure a certain number of pods are
running
○ Container (a docker container)
19. Magnum Resource Bay (K8S Cluster)
Source: https://www.openstack.org/summit/vancouver-2015/summit-videos/presentation/magnum-containers-as-a-service-for-openstack
20. Magnum Resource Bay (Swarm Cluster)
Source: https://www.openstack.org/summit/vancouver-2015/summit-videos/presentation/magnum-containers-as-a-service-for-openstack
21. Magnum Roadmap ... (?)
● Ironic Support
● Autoscaling
● Auto-Restarts
● Support for RKT
● RancherOS Support (?)
28. Magnum has the potential ...
● To become a Magnum Opus
● To provide cloud ready apps and services
● To manage Multi Tenant aware Container
Topologies
● To use OpenStack as a first class citizen for
container technology
● To help people helping people
29. About Kolla
● Containers for OpenStack
o Stackforge Project
o Docker Container based OpenStack
deployment (image based)
o Containerizing OpenStack Services
o Solves manageability and availability
problems
o Offers atomic upgrade of a running
system with minimal interruption in service
(10 seconds)