PinnedFeruz Urazaliev·Sep 25, 2025ADF vs. Databricks: Roles, Boundaries, and the Handoffs That Keep Platforms StableEnterprises get into trouble when they treat orchestration and transformation as the same job. They aren’t. One is about negotiating with…
PinnedFeruz Urazaliev·Aug 19, 2024Unlocking Databricks: 10 Hidden Features Every Data Engineer Should KnowDatabricks is a powerful cloud-based data platform that integrates seamlessly with Apache Spark for big data processing, machine learning…A response icon2A response icon2
PinnedFeruz Urazaliev·Jul 25, 2024Mastering PySpark: Advanced Techniques for Data EngineeringDive into advanced techniques for mastering PySpark, a powerful tool for data engineering. This guide explores sophisticated methods to…
PinnedInTowards DevbyFeruz Urazaliev·Jun 12, 2024What is Liquid Clustering in Databricks?Liquid clustering is a feature in Databricks that optimizes the storage and retrieval of data in a distributed environment. It’s designed…A response icon1A response icon1
Feruz Urazaliev·Sep 3Spark Says VARIANT Is GA. The Source Still Says Unstable.Shredding is on by default now. It rewrites your Parquet layout, changes write-path memory, and has no migration note anywhere.
Feruz Urazaliev·Sep 3I Fact-Checked Four Spark Features. All Four Release Notes Were Wrong.Not lies. Something more useful to understand: six ways a release note drifts from what actually shipped, and how to check in ten minutes.
Feruz Urazaliev·Aug 25Spark 4.2 Has Geospatial Types Now. It Has Five Functions.GEOMETRY and GEOGRAPHY are on by default. There is no distance function, no spatial join, and no way to reproject a coordinate.
Feruz Urazaliev·Aug 24Hyatt Cut Its Cheapest Award Night by 14%. In October It Will Cost You 14% More.Two 2026 changes stack, and for Sapphire Preferred holders every single cell on the award chart got more expensive.
Feruz Urazaliev·Aug 24Spark Declarative Pipelines, and the CDC Feature Nobody DocumentedSDP landed in 4.1 with a CLI and a YAML spec. Auto CDC arrived in 4.2 with working code, no documentation, and SCD Type 1 only.
Feruz Urazaliev·Aug 21Spark Shipped Millisecond Streaming and Forgot to Document ItReal-Time Mode landed in 4.1. The user guide 404s, the latency claims contradict each other, and Kafka is the only source that works.