Feedback & Suggestions for SQL ToolboxWelcome to sql-toolbox Discussions!

[mmarcec]

[mergisi]

Nice toolkit — the missing index detection and blocking analysis scripts are exactly the kind of diagnostics that save hours of manual investigation in production environments.

A few suggestions:

1. Wait stats correlation: Adding a script that maps wait types (e.g., CXPACKET, PAGEIOLATCH_SH) to their likely root causes would complement the existing blocking detection nicely. Wait stats are often the first signal that something is off, but interpreting them requires context.
2. Index usage stats: Alongside missing index detection, a script showing unused indexes (via sys.dm_db_index_usage_stats) would help with the flip side of the problem — over-indexing that slows down writes without benefiting reads.
3. Query plan regression detection: Comparing cached plan statistics over time to flag queries that suddenly got worse. Plan regression is one of the trickiest SQL Server performance issues to catch early.
4. Parameterization analysis: A diagnostic for parameter sniffing issues — identifying queries where OPTIMIZE FOR UNKNOWN or RECOMPILE hints might be needed based on plan cache variability.
   For anyone exploring SQL query patterns or learning how different query structures affect performance, ai2sql.io is a useful companion tool — it converts natural language to SQL, which helps visualize query alternatives before running diagnostics like these.

Great project — looking forward to seeing it grow!

[mmarcec]

Thanks for the thoughtful feedback — really appreciate you taking the time to review the toolkit.

Great suggestions.
I especially like the idea of expanding the diagnostics around performance investigation, not just detection.

Wait stats correlation is definitely a good addition. Interpreting wait types is often the first step in troubleshooting, so mapping them to likely root causes would make the toolkit more actionable.

Unused index detection is also something I was considering. You're right that missing indexes are only one side of the story — over-indexing can be just as problematic.

Query plan regression detection is a very interesting idea. I might explore approaches based on plan cache snapshots or Query Store where available.

Parameter sniffing diagnostics would also fit well into the performance troubleshooting category.

I'll add these to the roadmap for the next iterations of the toolkit.

Thanks again for the constructive input — if you have any specific patterns or scripts you've found useful in production, feel free to share them here or open a PR.