Some Lessons We Learned from Moving E-business Giant to Cloud Native
summary
Alibaba migrates legacy apps onto Kubernetes and proves production scale: nearly a million containers across ~100 datacenters and over a million containers in some counts, driving new tooling. Engineering eliminates "rich" app images in favor of lightweight sidecars and an admission-based sidecar injection service, implements an in-place update rolling strategy to avoid pod recreation, and builds a batch allocation planner to pre-place hundreds of thousands of pods for peak events—these reduce scheduler storms and stabilize upgrades at scale.
AI-written summary · about
chapters (18)
- 0:00Start
- 0:38Alibaba's Journey to Cloud Native
- 2:23Current State
- 3:19Architecture
- 10:23Eliminate "Rich Container"
- 18:15Cloud Native App Mgmt
- 23:28Workload Mgmt in Web-scale is Challenging
- 24:39Batch placement
- 29:32Application upgrade
- 30:02Introduce: in-place update
- 32:13Comparation
- 35:25Sidecar management
- 38:40SidecarSet injection
- 39:38What can SidecarSet do
- 41:00OpenKruise
- 42:38AdvancedStatefulSet
- 43:39SidecarSet and BroadcastJob
- 44:06Scalability Matters in Alibaba
related talks
44mSome Lessons We Learned from Moving E-business Giant to Cloud Native
22mKeynote: Cloud Native at Alibaba Scale
06mKeynote: Kubernetes is Now, Cloud Native is Future – Ding Yu
29mFrom EC2 to Kubernetes Takeaways from a Brownfield Migration
06mKeynote: The Cloud Native Ecosystem in China
08mKeynote: Cloud Native in the Enterprise: Production, Best Practice and Innovation
33mNavigating the Seas of Transformation: Migrating from EC2 to Kubernetes
30mMigrating a monolith to Cloud-Native and the stumbling blocks