There are many ways to configure your application inside a container. We will discuss the advantages and disadvantages of the different configuration possibilities. Also we will have a broad overview how some container environments like Kubernetes, Mesos (Marathon) and Swarm support you configuring your application. As a key take-away you will know the different configuration possibilities and be able to decide which one fits best for different application types. Event: inovex Meetup Köln, 08.06.2016 Speaker: Johannes M. Scheuermann, inovex GmbH Weitere Tech-Vorträge: https://www.inovex.de/de/content-pool/vortraege/

1. Application configuration in containers
Johannes M. Scheuermann
Köln, 08.06.2016

2. Johannes M. Scheuermann
IT Engineering & Operations @ inovex
〉 Software-Defined Datacenters
〉 Container ecosystem
〉 Cloud technologies
〉 High Availability & Scalability
〉 @johscheuer
〉 https://github.com/johscheuer/Brownbags/tree/ma
ster/docker_app_config
2

3. What about you?

4. Applicationconfiguration
4
Configuration
Static Dynamic
Image Mount/Env K/V store

5. • Inflexible
• Restart required
• Easy to setup
Static configuration
5

6. • Good for default configuration
• Dev / Prod parity
• Easy distribution
• Needs a rebuilt on change
Image (on build)
6

7. • Mount from host
• Inject environment variable
• Only simplekey/values
• Fixed from start time
• echo/sed in configuration files
• Can overwritedefault configuration
Start up time
7

8. • Flexible
• Application must support it
• Complexity
• Configuration management
• Key/Value store
Dynamic configuration
8

9. • Mount config from host
• Configuration management
• Complex configuration
• Independent of Container
Mount
9

10. • Very flexible
• Environmental variables
• Configuration files
• Templatefiles
• Separatedsetup needed
• Confd / Consul-template / envconsul
Key valuestore
10

11. • Kubernetes
• Configmap
• Secrets
• Mesos / Marathon
• Artifact Store?
• Swarm ?
• Environment variables
Environments
11

12. • Not all solutions will work
• Mount doesn’t work
• Config on all Nodes
• Key Value store
• Separated setup
• Additional complexity
Distributed
12

13. • Many backends
• TOML (Tom's Obvious, Minimal Language)
• Polling/Watch/Onetime
• CMD to run after changes
Confd
13

14. Confd (high level)
14
Host
etcd cluster
Prod
Test
prod
test
keys/prod/..
keys/test/..

15. Demo

16. • Similar to confd
• Only Consul
• HCL (HashiCorp Configuration Language)
• Supports plugins
• Rich feature set
Consul template/env
16

17. • Find the right choice
• What fits your application best?
• Combinations possible
• See example
• No holy grail
Conclusion
17

18. 18
We are hiring!
www.inovexperts.com

19. Q&A

20. Johannes M. Scheuermann
inovex GmbH
Johannes.scheuermann@inovex.de
CC BY-NC-ND inovex.de +JohannesScheuermann
github.com/johscheu
er
@johscheuer youtube.com/inovexGmb
H