Field notes on autonomous data.
Guides, product news, and perspective on moving data teams from maintenance to ownership.
What is a data product?
A dataset built like software — owned, tested, documented, and governed — so the org can trust it and build on it.
Read →Change Data Capture (CDC), explained
What CDC is, why it beats batch, and the methods — log-based, query-based, trigger-based — with their trade-offs.
Read →ETL vs ELT: what changed, which to use
Why ELT overtook ETL in the cloud era, when ETL still fits, and a practical way to decide.
Read →Replicate Postgres to BigQuery in real time
The approaches, the hard parts — schema drift, type mapping, deletes, backfill — and a reference setup.
Read →Log-based vs query-based CDC
The three CDC methods compared on source impact, completeness, and latency — and when each wins.
Read →Exactly-once delivery in data pipelines
At-most, at-least, exactly-once — what each guarantee means, why exactly-once is hard, and how systems get there.
Read →Schema drift: what it is, how to handle it
Why source schema changes break pipelines — and the strategies that keep data flowing safely.
Read →What is reverse ETL?
Syncing modeled warehouse data back into the operational tools business teams live in — data activation, explained.
Read →Data pipeline pricing explained
MAR, rows moved, connectors, credits — how the models work and how to estimate your real bill.
Read →Migrating off Informatica: a practical path
A structured, lower-risk alternative to a multi-quarter GSI engagement — inventory, convert, validate, cut over.
Read →