Stormdriver One Year Later
summary
Stormdriver is a Clouddriver aggregator written in Go that shards accounts across Cloud Driver instances, cutting per-instance memory roughly by a factor of 10–15 and keeping footprint for thousands of accounts under 100 MB. It implements a gRPC agent–controller tunnel so remote Cloud Driver instances run in customer clusters with local IAM, enabling secure remote deployments, self‑service onboarding, and stronger multi‑tenancy; deployed at scale and open sourced, it reports successful reductions in resource use and manageable operational tradeoffs.
AI-written summary · about
chapters (12)
- 0:00Introduction
- 0:21Cloud Driver problems motivation
- 1:03Desired features account sharding
- 2:34Cloud Driver limitations architecture
- 3:38Storm Driver design deployment
- 4:42Configuring Storm Driver Spinnaker
- 5:49Controller-Agent model remote deployments
- 7:56Use cases multi-cloud deployments
- 9:23System architecture overview
- 11:35Lessons from one year deployment
- 13:23Open source release multi-tenancy
- 14:48Q&A
related talks
15mSharding Clouddriver with Stormdriver
29mScaling Spinnaker for multi-region multi-cloud SaaS deployments
28mCloud in your Cloud
38mHighload — An Unexpected Journey
20mLow latency scalable web crawling on Apache Storm #bbuzz
37mStorm, a Distributed Real-time Data Processing System
50mWeathering the Cloud Storm: Building Resilient Geo-Distributed Apps
27mEmpowering Resilient Scalability: The Power of Choreography in Distributed Systems