Building Event Collection SDKs and Data Models
summary
Describes building and shipping 20+ SDKs for web, mobile, IoT and server use, all grounded by JSON Schema validation so events are strongly typed and versioned. Core pieces are emitters, subjects and client-side sessionization, with retrying event stores on mobile and configurable server trackers. On the warehouse side models are incrementalized, aggregating heartbeats into page_views, sessions and user tables to avoid full recomputes and make queries fast and cheap.
AI-written summary · about
chapters (10)
- 0:00Snowplow and event collection
- 3:57Snowplow trackers and SDKs
- 10:35Tracking SDKs and schema events
- 17:18Handcrafted vs auto SDKs
- 19:26Server vs client SDKs
- 24:00SDK defaults simplification
- 25:03Raw event data to models
- 30:33Data aggregation for analytics
- 34:22Efficient incremental models
- 37:55Key takeaways
related talks
38mBest Practices to Design & Build Event-driven Applications
26mAdobe Experience Platform Data Collection
11mEvent Modelling
09mDeveloping and supporting 12 SDKs with a team of three
52mEvent Sourcing for the Cloud Developer
54mThe Data Side of Event Driven Architecture
45mThe Future of Data in Ember
22mSDKs, Relations, Love & Thunder