Performance diagnostics in Yandex StoreDoc
Yandex StoreDoc provides a built-in tool for diagnosing your database cluster performance. You can use it to analyze Yandex StoreDoc profiling and index metrics.
Warning
This feature is in Preview.
Enabling statistics collection
To use this diagnostic tool, enable statistics collection during cluster creation or update. By default, the system only collects statistics for queries longer than 300 ms. If necessary, change the slow query collection threshold by adjusting the slowOpThreshold setting in your database configuration.
Getting profiling data
For read, write, and other database queries, such as aggregate operations, you can view the data collected by the Yandex StoreDoc profiler:
- Chart with metrics for the selected data slice. You can toggle individual categories in the chart by clicking the category name in the chart legend.
- Table with statistics broken down by query type.
To get this data:
-
Open the folder dashboard
. -
Navigate to the Yandex StoreDoc service.
-
Click the name of your cluster and select Performance diagnostics → Profiling.
-
Specify a time range and adjust filters as needed.
-
Select:
- Data segment:
FORM: By query typeHOSTNAME: By shard’s primary hostNAMESPACE: By collection nameSHARD: By shardUSER: By user
- Data grouping:
COUNT: Number of queries.DOCUMENTS_EXAMINED: Number of collection documents scanned per query.DOCUMENTS_RETURNED: Number of returned documents.DURATION: Query execution duration.KEYS_EXAMINED: Number of index keys scanned per query.RESPONSE_LENGTH: Size of the document returned by the query.
- Aggregation function for the selected metric:
AVG: AverageSUM: Sum
- Data segment:
Getting information about indexes
The system provides collection access statistics and indexing recommendations for specific fields to improve query speed.
- Open the folder dashboard
. - Navigate to the Yandex StoreDoc service.
- Click the name of your cluster and select Performance diagnostics → Suggested indexes.
- Specify a time range and adjust filters as needed.
The table shows recommended index fields and collection access counts.