Going MultiCloud
summary
Dynatrace expands from AWS into Azure and GCP and finds the platform effort rises sharply — expect 70% extra platform work to script around vendor deviations. Storage and network throughput vary wildly, sometimes up to 70% deviation, and list prices differ ~25% pricing differences. Solution: a unified internal developer platform, Kubernetes (Rancher), a self-hosted Kafka with Strimzi instead of three managed Kafka offerings, vendor-managed Postgres where acceptable, strict golden-path CI/CD, and single-pane observability for cross-cloud comparisons.
AI-written summary · about
chapters (25)
- 0:00Introduction
- 1:07Who uses multicloud today
- 2:12Multicloud myths
- 3:15Why adopt multicloud
- 4:18Dev and staging in multicloud
- 6:08Tooling and delivery
- 7:34Dino TRS’s approach
- 8:19Impact on teams
- 9:21Cloud vendor differences
- 10:04Pricing considerations
- 11:12Machine performance
- 12:34Networking challenges
- 14:01Specialized machines
- 15:03Database upgrades
- 16:50PostgreSQL differences
- 18:58Kafka hosting
- 20:46Security gaps
- 21:08Unified delivery platform
- 22:13Disaster recovery
- 22:35Toolset simplification
- 24:24Platform architecture
- 26:10Observability
- 27:57Summary
- 28:20Q&A patching
- 30:45Q&A database choices
related talks
34mCloudy With a Chance of Kubernetes: Going From One To Three Cloud Providers
34mLessons Learned from Five Years of Multi-Cloud
28mFlexible Cloud, Whatever The Workload
40mPanel Discussion: Going Multi-Cloud for Realz. Stories From Media, Retail, & Infrastructure
20mDesigning for Multicloud
11mCloud confused - differences between public cloud vendors
22mMulticloud by Design
15mJourney to Hybrid Multi-Cloud