skinny.

Latest / Deep End Era. Engineering with Sankalp

Replication Topologies, Logical Decoding, Change Data Capture (CDC), Split-Brain Fencing, and DBSP Incremental View Maintenance

How do you stream every row change out of a primary transactional database into search indexes, caches, and real-time materialized views without ever missing a commit, corrupting an initial snapshot, suffering a split-brain primary during network partitions, or recomputing multi-way SQL joins from scratch? In the Series Six finale, we trace physical vs logical replication, PostgreSQL Logical Decoding and Replication Slots, Debezium/DBLog lock-free consistent snapshots, the Transactional Outbox pattern, Split-Brain Fencing Tokens, and the Z-Set algebra of DBSP (VLDB 2023) powering Materialize…

The skinny

The skinny isn't ready yet — notes appear once the transcript is processed.