Beyond traditional CI
by Gildas Cambridge
1. Machine Mgt
1.1. Manage config
1.2. Manage binaries
1.3. golden vm does not exist
1.3.1. VM Appliance is start
1.3.2. stemcell vm's (cfr. puppet guy)
1.4. Or rebuild from scratch ?
2. examples for different environments
2.1. webapp in opensource environment
2.1.1. cruise/hudson?
2.1.2. cargomaven/capistrano/vlad the deployer?
2.1.3. cfengine/puppet/chef?
2.2. backoffice software in proprietary environment
2.2.1. MS SMS
2.2.2. sysprep/WDS
2.3. virtual and physical
2.3.1. xen VM + deployment tools (EC2, openqrm)
2.3.2. fai+pxe booting+expect script+cfengine
3. audience
3.1. integrators
3.2. qa people
3.3. ops people
3.4. dev interested in "what happens next"
4. System Source definition = 2 parts
4.1. OS packages + Other software
4.2. Configuration of each component
5. Changes
5.1. Patches
5.1.1. WSUS
5.1.2. RPM updates
5.2. Config changes
5.3. Software changes
6. scope of working software = works in ops
6.1. possibly including monitoring
7. traditional CI stops at software building and testing
7.1. add deployment, performance testing, integration
8. different pipelines need to be integrated
8.1. application
8.2. environment
8.2.1. Server
8.2.2. Network
8.2.3. Desktop
8.3. data
8.3.1. Migration of data to new instance
9. Monitoring
9.1. Integration Monitoring
9.2. Unit Monitoring
9.3. Regression Monitoring
10. Any ideas for applying the same for Network?
10.1. Virtual switch?...
10.1.1. either virtual or real will work
10.1.1.1. scripting needed
10.2. adding network tests (ping, routing?)?
11. evolving system administration
11.1. Started with manual installation of everything
11.2. Then we automated the installation and kept doing maintenance on the installation
11.3. Next step rebuilding the whole environment
11.3.1. Special attention to data migration