With growing demand for containers in the enterprise, build pipelines are a bottleneck to success. Traditional workflows can't release application candidates quickly enough to fulfill demand. With over 400 development teams across many different business units, Pearson had to move away from massive installs of traditional build pipeline tools and rethink the entire concept. In this talk we'll demonstrate how we have built in security compliance, performance testing, quality assurance, abstracted away complexity, reduced overhead, aim to recover 10% of developers time and turned build tools into cattle.
This represents the story to date of an in-flight engineering project to modernise the digital estate of a global enterprise organisation and how scale of the operation is leading us to challenge many established beliefs. Attendees will walk away with everything from workflows to code which they can use to get started in their own endeavors.
Sched Link:
2. 2
We cannot solve our problems
with the same thinking we used
when we created them.
Albert Einstein
Education is what remains
after one has forgotten
everything he learned
in school.
Albert Einstein
4. 4
Principal Systems Architect
Technical Lead on Project Bitesize
Pearson's Enterprise Platform-as-a-Service based on
Kubernetes
Michael Ward
www.devoperandi.com
@devoperandi
7. 7
Pearson is no small potatoes:
36,000 servers
2,000
applications
400+ Dev
Teams
Up to 30 Micro-
services per
application
1:1 build
server to Dev
team
16. 16
"Well we spent 6 months solid building a good
starting point"
Pearson Developer
17. 17
"we never upgrade Jenkins once stable cause
we can't get time for it"
Pearson Developer
18. Stakeholder Requirements
Reduce Migration Costs Compliance
"pre-baked security"
Standardization
"same build process across whole Pearson estate"
Visibility
Performance Testing
"automated perf testing"
Ease of Use
"if you don't get this I can't help you"
< 5% Time required to Manage Quality Testing
18
19. Our Requirements
Geographically Distributed
"Abstract Deployment but allow it anywhere"
Cattle
"Jenkins as Cattle"
Fast
"< 1hr"
Automated Upgrades
Self Configuring Scalability
"Horizontal"
Feedback Loops
"Clear communication across teams"
19
22. Key Concept:
22
1 dev team to many builds
Dev Team
Namespace Build Servers
console-dev
console-stage
console-dev
console-prod
console-stage
console-prod
console-prod
console-stage
console-prod
Oregon
Ireland
Private DC
43. 43
Meet the Team
Director PM Engineer
Chris Jackson Mark Stocker Simas Cepaitis
Engineer
Bill JorgensenJohn Shirley Martin Devlin
Engineer Engineer
Peter Idah
Engineer
46. 46
PaaS as an ecosystem
Build
API Management
Event Driven
Automation
Runtime
Monitoring
Secrets
Orchestration
App Config
47. Key Concept:
47
1 dev team to many namespaces
Dev Team
appeng-dev
appeng-stage
appeng-prod
appeng-stage
appeng-prod
appeng-prod
Namespace
Kubernetes
ClusterProject
console-core
console-ui
console-app