How much open source is in cloud services?
summary
Reverse engineering public repos shows managed OpenShift services stitch together upstream pieces into a reproducible pipeline: cluster provisioning via Hive operator + OpenShift installer, multi‑cloud lifecycle via ACM/ACM (ACM downstream), configs shipped with syncsets and a configure-alertmanager-operator, monitoring with Prometheus/Alertmanager, Splunk forwarders, and runbooks. SRE automation runs thousands of clusters in ~30–40 minutes deploys, adds ~20–30 extra operators, and CI uses TestGrid/Prow with Ginkgo tests (one asserts 50 image streams); much is open source, but some operational glue remains closed.
AI-written summary · about
chapters (16)
- 0:00Start
- 6:07Full Stack Management via Site Reliability Engineering (SRE)
- 7:23Lifecycle of a Managed Cloud Services Cluster
- 9:42Hive to the rescue!
- 10:10Hive architecture
- 10:51Open Cluster Management
- 11:44TLS ENCRYPTION
- 13:40Managed Cluster Config
- 15:24Validating Admission Hooks
- 16:36Platform Monitoring
- 18:44Configure Alertmanager Operator
- 21:02What is monitored?
- 23:44Logging
- 25:09How SRES Access to the Clusters?
- 28:55k8s tests
- 30:37Inspirational resources
related talks
47mOpen Source Ops Fast Track with Operate First
51mMaking Complex Open Source Safe for Operations
25mOpen source observability for private cloud: mission impossible or not?
42mOpen Source in the Age of Cloud and Microservices
21mBack to the original mission: an open cloud operating system
38mOS-Climate & SIG-SRE: Benefiting from the Operate First Community Cloud
48mKeynote: A year in the life of an Open Source SRE
28mOpen Source at Scale Building Operation in a large company