Designing Real Schemas

Application

Apply MergeTree fundamentals to real schema-design decisions for events, metrics, and log workloads.

What this module covers

Events table schema pattern
Metrics and time-series table schema pattern
Kafka table engine
Materialized views (insert-triggered, not query-time)
Denormalization vs. JOINs
Nested and Array columns for semi-structured data
ARRAY JOIN
Window functions for sessionization and time-series
TTL for data retention
Compression codecs (choosing by column type)
External dictionaries for enrichment lookups
Projections (first mention)

Lessons