Grading queries and analyzing logs in QueryGrade

QueryGrade works at two scales: one query at a time, and across a whole log of queries.

A single query

Paste the query, optionally choose the database dialect, and submit. QueryGrade runs it through a set of specialized analyzers - covering SELECT lists, joins, WHERE clauses, ORDER BY / GROUP BY, indexing, and subqueries - and combines their findings into a grade, a score, and a ranked list of issues.

Common patterns it catches include SELECT *, functions applied to columns in a WHERE clause, leading-wildcard LIKE '%x', cartesian and excessive joins, ORDER BY RAND(), and likely-missing indexes. Each finding comes with a concrete suggestion.

A log of queries

If you already have a database running, export a MySQL slow query log or general query log and upload it. QueryGrade parses the log and uses anomaly detection to surface the queries that stand out as unusual or expensive, with a score per query, so you can start with the worst offenders instead of reading the whole file.

Making the most of it

• Start with a query you already suspect is slow - the feedback is the fastest way to see how QueryGrade thinks.
• For log analysis, a representative window of real traffic gives better anomaly signal than a tiny sample.
• Sign in so your graded queries and feedback are saved and can inform later results.

Want a second opinion on a result? Contact support and we’ll take a look.