Posts
Browse all articles.
-
Data This Week #20
5 min readKafka's nine-layer architecture breakdown, self-healing pipeline barriers, ClickHouse full-text search on object storage, Google Cloud Next '26 data infrastructure, watsonx.data semantic layer, Neo4j on Databricks, DuckDB internals, and handling messy Excel ETL.
-
Data This Week #19
5 min readSlack's SSH-to-REST EMR migration, clusterless Iceberg Lakehouse with DuckDB, Iceberg v4 metadata proposals, Spark 4.0 on EMR GA, Apache Gravitino unified catalog, and Databricks Omnigent for AI agent orchestration.
-
Data This Week #18
4 min readSupabase Multigres for Postgres scaling, Netflix's dynamic Cassandra partition splitting, dbt Core v2 on Rust, SQL/PGQ graph queries in Postgres 19, and IAM fundamentals for data engineers.
-
Data This Week #17
4 min readPulsar 5.0 Scalable Topics, Iceberg multi-engine query routing, Netflix's high-throughput graph abstraction, Iceberg partition evolution, AWS Glue automation, and community thoughts on Databricks' BI migration tool.
-
Data This Week #16
5 min readDatabricks as a unified platform, PostgreSQL pluggable storage engines, Iceberg 1.11.0 highlights, database egress costs, JDBC query caching, and dbt column-level lineage.
-
Data This Week #15
6 min readSpark Declarative Pipelines for financial lakehouses, ten AWS Glue & Iceberg fixes, MOR as an architectural shift, DuckDB's Quack protocol, SQL fraud patterns, Kafka checkpoint patterns, and the LLM-for-validation debate.
-
Data This Week #14
6 min readFlink CDC streaming ELT from MySQL to Kafka, the LLM engineer's stack map, Ursa's diskless Kafka fork, Iceberg write mechanics, Instacart's billion-product search, Jikkou 1.0, and the AI knowledge-base debate.
-
Data This Week #13
5 min readSpark memory tuning, row-level validation tiers, Postgres RLS pitfalls, Stripe's sharding at 5M QPS, Aurora DSQL vs Postgres, Velero joins CNCF, and SQLGlot 5x faster with mypyc.