Différences de recommandations dans Azure Advisor

Azure Advisor is the single recommendation engine, but its views are scoped differently, which explains the behavior.

Key points from the product behavior described in the documentation:

1. Advisor works at subscription scope
   • The Advisor dashboard aggregates and shows personalized recommendations for all subscriptions and all supported resource types in one place, grouped by categories such as Cost, Reliability, Performance, Security, and Operational excellence.
   • Filters on the Advisor dashboard (subscription, resource type, category) control what is visible there, independently of what each individual resource blade shows.
2. Resource “Recommendations” tabs are scoped to that specific resource
   • When opening the Recommendations tab on a VM (or any other resource), only recommendations that Advisor associates directly with that resource are shown.
   • Some Advisor recommendations are subscription‑level, resource‑group‑level, or cross‑resource (for example, a recommendation that compares usage across multiple VMs, or that comes from another service such as Microsoft Defender for Cloud). These can appear in the Advisor dashboard but not on an individual VM’s Recommendations tab.
3. Different surfaces can expose different subsets
   • The documentation explicitly notes that not all recommendations for a service are surfaced in every UI surface. For example, Azure Cosmos DB recommendations can appear either in the service pane or in Advisor, and “not all recommendations shown in the Azure Cosmos DB pane are available in the Azure Advisor and vice versa.” The same pattern applies more broadly: some recommendations are only visible in Advisor, some only on the resource blade, and some in both.
4. Reliability of the sources
   • Both the Advisor portal and the resource’s Recommendations tab read from the same underlying recommendation engine; neither is “more correct.”
   • For automation and complete coverage, the Advisor API and the Advisor dashboard are the authoritative sources because they operate at subscription scope and include cross‑resource and subscription‑level recommendations that a single resource blade cannot show.
   • The resource’s Recommendations tab is best viewed as a convenience view focused on that one resource, not as a complete list of everything Advisor knows about that resource’s environment.

In practice, for automation and reporting, rely on the Azure Advisor API (and validate against the Advisor dashboard). Use the resource‑level Recommendations tab only for quick, per‑resource troubleshooting.

References:

• Introduction to Azure Advisor
• Automated recommendations for Azure Cosmos DB
• Operational Excellence recommendations
• Reliability recommendations
• Performance recommendations