# Databricks Field Guide > An opinionated guide to building, running and paying for Databricks, written by the > engineers at TechFabric. Free, not gated, and no sign-up. Written by the engineers at TechFabric, a Databricks partner with 80 Databricks-certified engineers inside a team of 115+. We migrate warehouses onto the lakehouse, design Unity Catalog governance, and build the applications and agents that run inside the workspace. The work is at https://www.techfabric.com/databricks. AI assists the drafting, and a TechFabric engineer reviews every page before it is published: how the guide is written, checked and corrected is at https://databricks.techfabric.com/about/. - [How we write this guide](https://databricks.techfabric.com/about/): who writes and reviews it, how AI assists the drafting, the fact-check and the corrections policy. Every chapter is also available as Markdown at https://databricks.techfabric.com/guide/.md, and the whole guide as one document at https://databricks.techfabric.com/llms-full.txt. ## Start here - [Databricks guide for engineers: where to start](https://databricks.techfabric.com/guide/): Where to start with Databricks for the job you have, the chapters that answer it, the first step we'd take, and how to use this free guide from an agent. - [Databricks vs Postgres and a dashboard: when you need more](https://databricks.techfabric.com/guide/not-just-postgres/): "Databricks vs Postgres for reporting: when Postgres and a dashboard is enough, the three signs it has run out, and how Lakebase changes the move." - [Databricks pricing: what a DBU costs and what decides the bill](https://databricks.techfabric.com/guide/what-it-costs/): What a DBU costs on each Databricks compute type in October 2026, DBUs an hour by warehouse and cluster size, model serving rates, and what cuts the bill. - [How to learn Databricks: a learning path and certification](https://databricks.techfabric.com/guide/how-to-learn-databricks/): "A Databricks learning path: what to build first on Free Edition, which free Academy pathway to follow, which certification to sit, and where to ask a person." ## Storing - [What is Unity Catalog? Databricks governance in practice](https://databricks.techfabric.com/guide/unity-catalog/): How Databricks Unity Catalog governs tables, files, models and agents, the catalog layout we'd use, row filters and ABAC in SQL, and where it stops. - [Delta tables on Databricks: storage, liquid clustering, features](https://databricks.techfabric.com/guide/tables-and-storage/): "How Databricks tables are stored: managed vs external, Delta vs Iceberg, liquid clustering, predictive optimization and deletion vectors, by runtime." - [Databricks Lakebase: managed Postgres with branching](https://databricks.techfabric.com/guide/lakebase/): What Databricks Lakebase is, what it costs, and how branching, restore, the Data API and synced tables behaved when we built an app on it live. - [Databricks LTAP explained: how it differs from HTAP and Zero ETL](https://databricks.techfabric.com/guide/ltap/): "Databricks LTAP explained: what lake transactional/analytical processing is, which Lakebase pieces ship today, how it differs from HTAP and Zero ETL." - [Medallion architecture on Databricks: bronze, silver and gold](https://databricks.techfabric.com/guide/medallion-architecture/): "Medallion architecture on Databricks: what bronze, silver and gold each promise, the Lakeflow pipelines code that enforces it, and when to use fewer layers." ## Processing - [Databricks change data capture (CDC) and real-time data](https://databricks.techfabric.com/guide/real-time-and-cdc/): "Change data capture on Databricks: Lakeflow Connect CDC connectors, AUTO CDC for SCD type 1 and 2, real-time mode, Zerobus, and picking a latency tier." - [Lakeflow Jobs: orchestrating Databricks jobs in production](https://databricks.techfabric.com/guide/orchestration/): "Lakeflow Jobs in production: triggers, for-each and if/else tasks, retries and timeouts, and the system table queries that audit a jobs estate." ## Analysing and serving - [Databricks metric views and Power BI: where definitions live](https://databricks.techfabric.com/guide/reporting/): How Databricks metric views work as a semantic layer, how Power BI reads them in DirectQuery, and the dashboard habits that keep reporting cheap. - [Databricks Genie: build a Genie Agent people trust](https://databricks.techfabric.com/guide/genie/): What Databricks Genie One, Genie Agents and Genie Code are, the warehouse a Genie Agent needs, what it costs from 2027, and how we'd build one people trust. - [Databricks row-level security for multi-tenant reporting](https://databricks.techfabric.com/guide/multi-tenant-reporting/): Row-level security for multi-tenant reporting on Databricks, with the SQL for row filters, ABAC, embedded dashboards and OAuth custom identity claims. - [Databricks Apps: what they are, cost, and when to use them](https://databricks.techfabric.com/guide/databricks-apps/): Databricks Apps for developers, covering the runtime, on-behalf-of-user auth and scopes, Lakebase and warehouse resources, deploys and the hourly cost. - [Delta Sharing, Databricks Clean Rooms and the Marketplace](https://databricks.techfabric.com/guide/sharing-and-collaboration/): Databricks Clean Rooms and Delta Sharing (OpenSharing) in practice, with the SQL to share a table, the clean room approval rules, and the audit queries. - [MLflow on Databricks: tracking, model registry and serving](https://databricks.techfabric.com/guide/mlflow/): Runnable MLflow 3 code for Databricks covering table lineage, the Unity Catalog model registry and aliases, deployment jobs, serving and GenAI tracing. ## Generative AI - [Databricks Foundation Model APIs and Unity Gateway](https://databricks.techfabric.com/guide/foundation-models/): How Databricks Foundation Model APIs work, which models are pay-per-token, what they cost in DBUs, and why every call goes through Unity Gateway. - [RAG on Databricks: AI Search and context engineering](https://databricks.techfabric.com/guide/context-and-retrieval/): RAG and context engineering on Databricks with AI Search (formerly Vector Search), chunking, hybrid search, reranking and filters by who is asking. - [Databricks AI agents: Agent Bricks, evals and governance](https://databricks.techfabric.com/guide/agents/): Building AI agents on Databricks with Genie Agents or the Agent Bricks CLI, scoring them with MLflow 3, and keeping them inside the caller's permissions. - [Governed actions for Databricks AI agents and write-back](https://databricks.techfabric.com/guide/governed-actions/): Unity Catalog governs who reads data, not what an AI agent may do about it. How to put a policy, evidence and durable execution between an agent and a write. ## Operating - [Databricks migration guide: from a warehouse to the lakehouse](https://databricks.techfabric.com/guide/migration/): How to migrate a warehouse to Databricks with Lakebridge and federation, which parts overrun, and how to prove parity before switching the old system off. - [Databricks cost management and performance tuning](https://databricks.techfabric.com/guide/cost-and-performance/): Databricks cost management and performance tuning: attribute spend with tags and system.billing SQL, pick the right compute, then fix slow queries in order. - [Databricks security: identity, network, keys and audit logs](https://databricks.techfabric.com/guide/security/): Databricks security in practice: the identity on each path to the data, ABAC policies, serverless egress, customer-managed keys and audit log SQL. - [Databricks CI/CD with bundles, Terraform and environments](https://databricks.techfabric.com/guide/ci-cd/): Databricks CI/CD with Declarative Automation Bundles (formerly Asset Bundles) and Terraform: a runnable databricks.yml, a GitHub Actions deploy and promotion. - [Databricks data observability: freshness, volume and quality](https://databricks.techfabric.com/guide/observability/): What data observability means on Databricks: freshness, volume and quality checks, system table SQL for job health, anomaly detection and alerts with owners. - [Databricks disaster recovery and high availability](https://databricks.techfabric.com/guide/resilience-and-recovery/): Databricks high availability vs disaster recovery: what the platform recovers, managed DR and its limits, Lakebase HA and cross-region DR, and a rehearsed plan. ## Reference - [Databricks vs AWS, Azure and Google Cloud data services](https://databricks.techfabric.com/guide/platform-comparison/): Databricks vs AWS, Azure and Google Cloud data services, service by service as of October 2026, where the native stack wins, and Databricks or Microsoft Fabric. - [Databricks vs Snowflake: cost, workloads, when to move](https://databricks.techfabric.com/guide/snowflake/): Databricks vs Snowflake on billing, storage, AI and Postgres, why moving an unchanged workload mostly moves the bill, and how to migrate when it's right. - [Delta Live Tables, Mosaic AI and other Databricks renames](https://databricks.techfabric.com/guide/renames/): Delta Live Tables is now Lakeflow pipelines. Where the DLT docs moved, and the other Databricks renames, with where each old name survives in code and billing. - [Databricks valuation, revenue and customers, by the numbers](https://databricks.techfabric.com/guide/databricks-by-the-numbers/): Databricks' valuation, revenue run-rate, customer counts and acquisitions, each traced to the release or filing that produced it, dated and linked. - [Databricks glossary: terms, old product names and meanings](https://databricks.techfabric.com/guide/glossary/): A Databricks glossary with one-sentence definitions of Unity Catalog, Lakeflow, Lakebase, Genie and AI terms, the old product names, and links to each chapter. ## Margin notes What Databricks did, and what we read into it. Dated, with a position. Feed: https://databricks.techfabric.com/notes/feed.xml - [MemCurrent on Databricks: what it is and when we'd use it](https://databricks.techfabric.com/notes/memcurrent-databricks-agent-memory/) (2026-10-21): MemCurrent is the design behind Databricks agent memory in Agent Bricks and Genie. How we would build agents on it, and where we would keep it out. - [Databricks AI Runtime is GA: training beside governed data](https://databricks.techfabric.com/notes/ai-runtime-is-generally-available/) (2026-10-21): AI Runtime, serverless GPUs for training and fine-tuning on Databricks, is generally available. Where we would train on it and the limits to plan for. - [Databricks Private Network Gateway: serverless into your network](https://databricks.techfabric.com/notes/databricks-private-network-gateway-serverless-reaches-private-networks/) (2026-10-13): Private Network Gateway gives Databricks serverless compute one connection into your own network. What it reaches, what it costs and where to pilot it. - [Databricks Feature Views are GA: what changes for feature code](https://databricks.techfabric.com/notes/feature-views-are-ga-and-the-point/) (2026-10-12): Databricks Feature Views reached general availability on 12 October 2026 for batch, streaming and request-time data. What to check before adopting them. - [Databricks LTAP Direct Writes in beta: faster synced table loads](https://databricks.techfabric.com/notes/ltap-direct-writes-is-the-first-piece/) (2026-09-13): LTAP Direct Writes loads synced tables straight to storage, off the app's endpoint. What it covers, and the latency test we'd run before trusting it. - [Databricks revenue growth: what the $7 billion run-rate means for you](https://databricks.techfabric.com/notes/the-growth-is-your-invoice/) (2026-09-12): Databricks run-rate grew over 80% while customers grew a third. The gap is existing customers spending more. If you are one of them, that growth is your bill. - [Databricks Lakebase: why Databricks bought a Postgres to serve apps](https://databricks.techfabric.com/notes/databricks-bought-a-postgres/) (2026-09-12): Databricks bought Neon in May 2025 and now headlines Lakebase in funding releases. The lakehouse needed a real database, and that says where apps are going. ## For coding agents: the MCP server The guide is also a public, read-only MCP server at https://databricks.techfabric.com/mcp, with no sign-in. Its tools are search_guide, get_chapter (a chapter or one section, code intact), get_code_examples, whats_new and list_chapters. Answers come from the published pages and change with each deploy, and a chapter, section or code example comes with the Databricks documentation links its page cites. Add it to Claude Code with `claude mcp add --transport http databricks-field-guide https://databricks.techfabric.com/mcp`. ## Working with TechFabric - [Databricks consulting](https://www.techfabric.com/databricks/consulting): what we are engaged for. - [Unity Catalog work](https://www.techfabric.com/databricks/unity-catalog): governance that survives an audit. - [Cost optimization](https://www.techfabric.com/databricks/cost-optimization): the spend that bought nothing. - [Migration Readiness Sprint](https://www.techfabric.com/databricks/migration-readiness): two weeks, fixed fee. - [Hire Databricks engineers](https://www.techfabric.com/databricks/hire-engineers): certified people, as a team. - [Talk to an engineer](https://databricks.techfabric.com/contact/). - [Events](https://databricks.techfabric.com/events/): the Gilbert Databricks User Group sessions TechFabric hosts, with recordings afterwards.