MergeTree Sort Key Design for Pre-Flattened Wide Tables
How to order columns in a MergeTree table to balance pruning and compression.
Callum Dzenga
The Probe
Denormalization and Join Avoidance Patterns
Projection Design for Alternate Sort Orders on Denormalized Tables
Narrow projections let denormalized tables serve multiple sort orders without full duplication.
Seren Almodovar
· 9 min read
What Is Stream Processing in Real-Time Analytics Pipelines
Stream processing transforms continuous events into queryable results for real-time analytics.
Callum Dzenga
The Feed
SLO vs SLI Instrumentation for High-Cardinality Event Streams
Choosing between metric and event-based SLOs determines what data survives to explain breaches.
Priya Venkatasubramanian
Distributed Traces Definition and Data Structure for Storage Systems
Callum Dzenga
Row-Oriented vs Column-Oriented Execution for Mixed Analytical Workloads
Priya Venkatasubramanian
Predicate Pushdown and Primary Key Skipping in MergeTree Queries
Tobias Nkemdirim
Denormalization Tradeoffs in BigQuery and Redshift Schemas
Callum Dzenga
Kappa Architecture Trade-Offs for Real-Time Analytical Pipelines
Simplicity comes with costs in reprocessing, late data, and query serving.
Priya Venkatasubramanian
P99 Latency Stability at Growing Data Volumes in Managed OLAP Databases
Seren Almodovar
Retry and Backoff Strategies for Agent-Issued Analytical Queries
Callum Dzenga
OLAP vs OLTP Architecture Fit for Real-Time Customer-Facing Products
Tobias Nkemdirim