Databricks Unity Catalog governance at Indra

Indra Renewable Technologies cut EV charging data sprawl by consolidating on Databricks, and Databricks Unity Catalog governance sits at the center of that shift. The UK smart charger maker moved from a fragmented Azure stack to a single governed platform, according to Databricks.

The case matters beyond one vendor story. EV chargers are streaming sensors, and operators need governed telemetry at scale without paying for five tools to do one job.

The starting point: too many tools, too much overhead

Indra's data lived across Cosmos DB, Synapse, Azure Data Lake Storage, Azure Functions, Power BI and other Azure services. Each use case grew its own pipeline.

Costs rose organically. Logic was duplicated. Governance was limited, with no single catalog or lineage view, and business teams waited on data teams for routine numbers.

Why a single platform mattered

Indra started with a parallel prototype migration from Synapse to Databricks. Once results met expectations, it expanded to monthly Delta tables and shifted operational load from Cosmos DB to Databricks.

Databricks Unity Catalog gave it central visibility and managed governance for all data objects in one place, versus gluing together single-purpose solutions. It is positioned as the industry's only universal catalog for data and AI, and Databricks has pushed it as open source to challenge rivals on interoperability.

Databricks is now the de facto lakehouse standard for this consolidation play. StartupHub.ai data shows Databricks scores 82/100, well ahead of the Azure stack it replaced at 3/100, which helps explain why teams move off Synapse despite migration friction.

For Indra, the next phase was making governed data usable beyond engineering. Automated AI/BI dashboards and Genie Agents gave marketing, operations and product direct access without new tickets to the data team.

Replacing fragile pipelines with a medallion architecture

The clearest win was the fleet data pipeline. It had relied on multiple Azure functions and separate stages to ingest telemetry and transactions and push outputs to reporting.

The new design uses a medallion architecture. Bronze holds raw data as the single source of truth, silver transforms with PySpark, and gold publishes business-ready tables for reporting and client delivery.

That removed three legacy Azure functions. One codebase now serves three clients on a daily schedule, with 60% to 70% performance improvement over the legacy functions.

The pattern mirrors common Synapse to Databricks migrations that use medallion bronze, silver, gold and Unity Catalog for governance. The difference here is operational: fewer moving parts, one pipeline to maintain, and governed lineage from ingestion to delivery.

Turning reporting into self-service

Monthly KPIs had lived in Excel, with some reporting in Power BI under licensing constraints. That kept decisions slow and created handoffs.

The new production pattern runs from governed gold tables cataloged through Unity Catalog and served through a serverless SQL warehouse. AI/BI dashboards refresh automatically on schedule and replaced manual trackers.

Genie One is embedded directly in the dashboard canvas. Users move from a fixed visual to a natural-language conversation with the underlying data without leaving the page.

Operational dashboards are specific. A Device Voltage Report monitors compliance across the in-field charger population. A Charger Uptime Analysis view showed 14,975 devices with 90.25% average uptime over a selected period. A Grid Voltage dashboard for Worcester adds geographic context for overvoltage and undervoltage hotspots.

Genie Agents extend that to conversational analysis for telemetry, device firmware, and users, devices and vehicles. Marketing can ask how many devices delivered power in a month. Engineering can check device counts by model or firmware version. Product can explore vehicle makes on the platform. Genie generates the SQL and supports follow-ups.

The numbers behind the shift

The Synapse to Databricks migration delivered 80-90% business cost savings and 90% storage cost savings. Per active user cost fell from £1.30 in January to 45p in May.

Query latency dropped from 24.4 seconds to 3.5 seconds. That changes how often teams actually query the data.

StartupHub.ai data shows Databricks raised $5B in Strategic Financing in 2026 at a $190B post-money valuation, which frames why it can invest in serverless SQL, Genie and Unity Catalog advancements while Azure retires older Synapse paths.

Why this matters for platform choices

This is a lakehouse versus warehouse decision in practice. Indra chose governance plus open formats over point tools, and Unity Catalog advancements now emphasize eliminating table format lock-in and adding business-user discovery.

Competitors are chasing the same governance layer. StartupHub.ai data scores Snowflake at 73/100, Firebolt at 65/100, VARADA at 55/100, Upsolver at 54/100 and AWS Glue at 21/100, all trailing Databricks at 82/100 on our platform rating. Snowflake's Horizon and open table formats like Iceberg are the direct counter, and Databricks open-sourcing Unity Catalog is its answer on interoperability.

For EV and energy operators, the lesson is standardize governance first, then scale streaming, Delta tables, Lakeflow pipelines and AI on the same foundation. Indra's advice mirrors that: stay curious, read the docs, and bring Databricks experts in early.

What data challenges do EV-charging companies face?

Data spans chargers, databases, cloud storage, BI tools and separate pipelines, which creates duplicated logic, fragmented telemetry and limited governance while costs rise. Indra faced exactly that before consolidation, with no single catalog or lineage view.

How does Databricks enable self-service EV analytics?

Governed gold tables feed AI/BI dashboards on serverless SQL, and Genie One lets users query telemetry, device, firmware and vehicle data in natural language from inside the dashboard. That replaced Excel trackers and lets routine questions resolve in minutes instead of days.

How can EV-charging companies prepare for streaming and AI?

Build reporting on a governed foundation today so streaming and AI do not add sprawl later. Indra's medallion layers and Unity Catalog lineage leave room for streaming ingestion, Delta tables and Lakeflow pipelines without adding separate stacks.

What does Databricks offer energy companies?

Databricks combines data engineering, SQL analytics, machine learning and AI on one platform, and the company positions it for smart-meter analytics, grid monitoring, EV charging, demand forecasting and regulatory reporting. Unity Catalog provides the shared governance that makes those workloads reusable across teams.

Indra did not add another dashboard tool. It removed three functions, one warehouse and a spreadsheet habit, and made governance the product.