The PlanetScale Insights Anomalies, introduces intelligent query monitoring to detect slower-than-expected database queries.
PlanetScale’s Insights Anomalies aims to streamline the evaluation of a database’s well-being and the resolution of issues, as outlined in the company’s blog post. The main objective is to provide a concise snapshot of the database’s status and facilitate a straightforward troubleshooting process.
PlanetScale emphasizes the significance of not just identifying anomalies within a database but also comprehending their underlying origins. Insights furnishes pertinent metrics for each anomaly, encompassing high-level query metrics like the number of rows read and written per second, resource utilization metrics for database components (such as CPU and disk usage), and details regarding backups and deploy requests that could potentially affect shared resources.
Insights meticulously logs and preserves accurate query counts for each query pattern in a database. This meticulous recording facilitates a comparison between the execution rates of individual query patterns and the overall health metrics of the database, allowing for the detection of highly correlated queries.
PlanetScale Insights offers users an in-dashboard tool for a comprehensive examination of all active queries running on their database. This tool enables the identification of queries displaying issues such as excessive frequency, prolonged execution times, large data returns, or errors.
Users can navigate a performance graph to precisely determine when a query was affected and, when applicable, view the associated Deploy Request. Additionally, the tool provides a list of all queries executed in the last 24 hours, sortable by metrics such as rows read and time per query, facilitating thorough analysis.
This integrated tool empowers users to diagnose query issues effortlessly, streamlining the optimization of individual queries without the need for extensive investigation. The Anomalies tab further notifies users of any active issues, flagging queries that are running significantly slower than expected.