dataMinds News Round up –
May 2025

Author: Siddhant Mene

In “Cheatsheet for Designing a Modern Application Data Stack,” Siddhant Mene presents a practical, high-level guide for architects and engineers building contemporary data platforms. The article outlines essential components-such as data ingestion, storage, processing, orchestration, and analytics-highlighting how to select the right technologies for each layer. Siddhant emphasizes cloud-native patterns, modular design, and the importance of scalability and maintainability. The cheatsheet serves as a quick reference for professionals aiming to architect robust, future-proof data stacks that support both operational and analytical workloads.

Author: Daniel Neumann

Daniel shows how to configure Fluent Bit v4 on AKS to use the cluster’s managed identity (system‑assigned or user‑assigned) for authenticating the Azure Data Explorer output plugin—no service‑principal secrets needed—by retrieving the kubelet identity GUID, granting it ingest permissions, setting managed_identity_client_id to SYSTEM (or GUID), and verifying logs flow into ADX.

Author: Chris Webb

Chris Webb demonstrates how to use Fabric Data Agents to automatically document Power BI semantic models by leveraging INFO.* functions. He shows how to create a dependency table in a secondary import‑mode model using EVALUATE INFO.CALCDEPENDENCY(). A Data Agent is then configured with AI instructions to interpret that metadata. The agent can generate human‑readable documentation, answer impact‑analysis questions, and even produce Mermaid diagrams of measure dependencies. This approach eliminates manual DMV scripting and enables dynamic, conversational exploration of model logic.

Author: Koen Verbeeck

Koen Verbeeck demonstrates how to use dbt’s macro functionality to build a fully featured date dimension in a Microsoft Fabric warehouse. He begins by installing the community‑maintained dbt_utils package and discovering that its built‑in date_spine macro isn’t compatible with T‑SQL’s restrictions. To address this, he authors a custom Jinja macro named my_date_spine that leverages T‑SQL’s GENERATE_SERIES and DATEDIFF functions to produce a continuous list of dates. He then embeds that macro in a dbt model to generate a table complete with typical date attributes—year, quarter, month, ISO week, and flags for first or last days—without performance penalties. This pattern shows how dbt macros can make complex SQL logic reusable, maintainable, and optimized for the Fabric environment.

Author: Sandeep Pawar

Sandeep Pawar built an interactive Microsoft Fabric CU calculator—using Claude 3.7 and Grok 3—to estimate Copilot and AI Function consumption based on documented assumptions. It clarifies that Copilot is billed as background jobs, AI Functions under Spark, and the full logic is open‑source on GitHub.