BMW Software Factory - winning strategies to maximize value
summary
BMW moves its CI core services to AWS and scales a software factory that runs up to 110,000 CI pipeline executions daily and peaks at 80,000 vCPUs in parallel. Builds run on fresh EC2 AMIs with per-image caches and about 230 EC2 instance types to match jobs. Adopting spot instances for ~30% of jobs yields 50–70% discounts, switching to Carpenter and Fleet APIs raises utilization and drives a reported CI throughput improvement of ~50%.
AI-written summary · about
chapters (9)
- 0:00Introduction and goals
- 3:11Why BMW needs massive build infrastructure
- 7:29Three focus areas: quality speed cost
- 13:11How CI integration works
- 15:18Need to move from on-prem to AWS
- 22:50Learnings and achievements together
- 30:59Optimizing GitHub Enterprise and Kubernetes
- 34:55Collaboration and future plans
- 38:49Next steps and Q&A
related talks
36mHow BMW stays on the fast track by building the Connected AI Platform for digital products
27mData-driven and cloud-native ecosystems in the automotive industry
57mCreating value from confluence of technology forces (PRT250)
14mPrivate Cloud, OpenStack, and the BMW Datacenter
31mRealizing Industrial AI at Scale with
34mJourney towards a software-driven company
12mModernize Manufacturing with SAP on
46mThe Software Factory