Microsoft announced general availability of the modern Query Editor for Azure SQL Database on October 22, 2026. It brings multiple query tabs, improved object discovery, data previews, and interactive results into the Azure portal. Microsoft announcement.
This is a useful improvement for a quick investigation. My operational rule would remain simple: a convenient workspace is where an engineer works, not automatically the durable record of what they approved or executed.
Treat Saved Queries as Browser-Local Working Material
Microsoft says open tabs and SQL text are restored during the browser session, while saved queries are stored in the browser profile. Persistence behavior.
I would therefore keep reviewed operational queries in the team's existing controlled repository or runbook. A local saved copy can make recurring work easier, but it should not become the only copy of a query another engineer may need.
Give reusable queries a clear purpose, owner, and expected output. If an investigation produces a useful query, review and retain it through the same process the team already uses for operational tooling.
Before handing an incident over, confirm that the next engineer can obtain the query without access to the original browser profile.
Reconfirm the Database Before Running Selected SQL
The updated experience supports multiple query tabs and running selected SQL. It also offers Microsoft Entra or SQL authentication. Workspace and connection options.
My proposed preflight would verify the intended environment, server, database, identity, and statement selection before execution. Multiple tabs help an investigation, but they also make an explicit target check worthwhile.
Start a new workflow with bounded, read-only checks. If a later step changes data or schema, move it through the normal approval process rather than treating it as a continuation of harmless exploration.
Record the exact statement used for a consequential action. The text left in a tab afterward may no longer describe what was selected and executed earlier.
A Preview Is a Starting Point, Not a Complete Dataset
Object Explorer can search by name and filter by object type. Table and view previews can display up to 1,000 rows. Object discovery and previews.
I would use a preview to understand a schema or inspect representative values, then write an explicit query for the investigation's real question.
Do not infer that missing preview rows prove missing business data. Document the relevant conditions and the result needed to support the conclusion.
When reviewing a deployment, define the expected invariant first. A check should demonstrate something specific about the release, not simply show that the table contains plausible-looking rows.
Keep Interactive Results and Exported Evidence Distinct
Microsoft lists grid sorting and filtering, copying, and exports to CSV, JSON, or Excel. Results capabilities.
My handover record would include the SQL, execution context, time, and any interactive filtering relevant to the interpretation. That lets another engineer distinguish the query result from the particular view someone chose to inspect.
Before exporting, decide whether the investigation needs raw rows at all. Prefer the minimum evidence required, and use an approved destination with appropriate access.
An easy export button does not decide who should receive the data. Apply the same handling rules used for output from desktop tools or automated jobs.
Preserve the Existing Connection Boundary
For a first evaluation, I would use an already approved connection path and identity. If the portal reports a connection problem, establish whether the cause is expected policy or an actual configuration error before changing access.
Do not widen a firewall rule merely to complete a user-interface test. Have the network or database owner review any proposed access change separately.
Keep the pilot's scope small: one approved database, a few reviewed queries, and a handover check that another engineer can reproduce.
Practical Cloud Engineer Takeaway
Keep durable operational queries outside a single browser profile.
Reconfirm the target and selected statement before execution.
Treat data previews as limited exploration.
Record interactive filters when they affect an interpretation.
Review exports and connection changes through existing controls.
Who Should Care?
Azure SQL operators, application support engineers, and release teams using the portal for lightweight database checks.
Bottom Line
The modern editor makes short investigations easier. Reliable operations still depend on a clear execution target, reproducible evidence, and a change record that survives the browser session.
No database connection, query execution, data export, firewall change, or database modification was performed for this article.
Sources
Stay radical, stay curious, and keep pushing the boundaries of what is possible in the cloud.
Chriz
Beyond Cloud with Chriz
Comments