Three months with Azure Data Manager for Energy: the gotchas Microsoft's docs won't tell you
TechPreacher · x · 2026-09-26
Microsoft's Forward Deployed Engineering team shares a hands-on retrospective after three months building a data platform on Azure Data Manager for Energy (Microsoft's managed OSDU service), with every finding verified on a live instance.
Key gotchas
- Provisioning is a one-shot decision: instance name (1-15 lowercase alphanumerics), region, SKU, data partition names and the networking model are effectively permanent; changing region means full redeploy plus data migration. Provisioning takes up to 2 hours (measured 1h43m) — don't cancel it.
- Regional capacity is finite: they needed 5 attempts across 3 regions before finding one with backend capacity. A failed ARM deployment still leaves a live resource in Failed state — delete it (Microsoft.OpenEnergyPlatform/energyServices) and confirm no lingering ME managed resource groups before retrying.
- Identity: each instance needs its own Entra app registration whose application id becomes the unchangeable data-plane instance id; app id + client secret is admin access to everything, so plan rotation and vaulting upfront.
The writeup also covers entitlements, schemas, OSDU Search behavior, ingestion via Azure Data Factory, and operations — the checklist they wish they'd had on day one.
More from Infra
- TSMC approves $60.7bn in capital appropriations in 91 days, betting big on advanced nodes — julsimon · 2026-09-26
- Amazon raises $62bn in dollar bonds alone in 2026, borrowing like a utility — julsimon · 2026-09-26
- 7-year lead time for substation transformers signals AI's power bottleneck — basedjensen · 2026-09-26
- LiveKit Acquires Loophole Labs to Cut Agent Startup Times Under 3 Seconds — HowDevelop · 2026-09-26
- Peak-hour OpenAI subscription use can burn 3-4x the quota, users find — Strange_Owl_6291 · 2026-09-26
- Anthropic and OpenAI list identical headline prices, but cache read differs 4x — julsimon · 2026-09-26