Blogs
War stories from the CDC trenches
Real failures, real fixes, and how Fabric avoids them.
Apr 4, 2026 · 5 min read
Zero records. Six hours. Why your Airbyte CDC sync says it's running when it isn't.
A Postgres CDC sync ran for 5 hours 56 minutes and moved nothing, while every dashboard said it was fine. Here's why, the three fixes, and how to skip the whole saga.
Abhisek OjhaFeb 21, 2026 · 4 min read
Small files, big bills: the silent tax on every CDC pipeline
Continuous replication writes lots of tiny files, and object storage bills you for every one. A short field guide, built on Yunus Alper Körükcü's excellent write-up, plus how Fabric handles it.
Ismael KhanOct 1, 2025 · 6 min read
One hundred tables, one config you can't guess
Turning a wide source database into Iceberg tables used to mean hand-composing an interdependent pile of Debezium config and running one JVM per pipeline. Here is what kept breaking, and how Fabric turns picking tables into a running pipeline.
Archit AgarwalSep 9, 2025 · 6 min read
Why one blank string can stop your CDC pipeline
A legacy application stored missing values as '' in columns that hold UUIDs. One of those strings stopped the pipeline, and recovering meant a full re-snapshot. Here is the anatomy of the crash, and how Fabric turns a bad row into a quarantine entry instead of an outage.
Abhishek BJul 25, 2025 · 5 min read
Why your idle Postgres database is quietly running out of disk
A database with almost no traffic can still fill its disk with WAL. Three usual suspects, the queries to find them, and why a heartbeat is the fix nobody tells you about.
Gunnar M