Skip to main content
Version: current [25.0.x]

Limits

This page details limits for a variety of Dremio resources on a per-deployment basis.

note

System limits are hard limits enforced by Dremio. Recommended limits are not enforced, but surpassing these limits can lead to cluster instability.

Limits are grouped according to Dremio components, as follows:

Catalog

ItemLimitLimit TypeAdditional Notes
Field size in records32 KBSystemPrevents excessive heap usage.
Leaf columns in a table6,400RecommendedSee Creating and Querying Wide Tables.
Sources250RecommendedAdding too many sources can impact performance in the Dremio console and the size of the KV store on disk.
Spaces500RecommendedExceeding this limit can cause degraded response times in the Dremio console.
Folders in a space250RecommendedAdding too many folders can impact performance in the Dremio console and the size of the KV store on disk.
Views in a folder500RecommendedAdding too many views can impact performance in the Dremio console and the size of the KV store on disk.
Size of uploaded Excel file (.xlsx)10 MBSystemMaintains coordinator stability.
Size of uploaded files (other formats)50 MBSystemMaintains coordinator stability.
JSON, CSV, or TXT files300,000RecommendedMaintains coordinator stability (metadata collection).
Scripts per user1,000SystemGuardrail to prevent excessive space usage in the KV store.
Starred datasets500RecommendedGuardrail to prevent excessive space usage in the KV store.
Pinned datasets500RecommendedGuardrail to prevent excessive space usage in the KV store.
Wikis and Labels5,000RecommendedGuardrail to prevent excessive space usage in the KV store.
Local roles1,000RecommendedGuardrail to prevent excessive space usage in the KV store.
Local users1,000RecommendedGuardrail to prevent excessive space usage in the KV store.
Catalog name length256 charactersRecommended256 characters is an industry standard (prevents excessive space usage in the KV store).
Script size10 MBSystemPrevents excessive heap usage.

Execution

ItemLimitLimit TypeAdditional Notes
Leaf columns that can be queried800SystemSee Creating and Querying Wide Tables.
Array size (Parquet)128SystemToo many elements in a LIST or ARRAY can exceed direct memory allocation.
Query records returned via console or REST API1,000,000SystemCan lead to excessive disk usage. Use ODBC or JDBC to see full result set.

Metadata

ItemLimitLimit TypeAdditional Notes
Number of CSV/JSON/TXT datasets promoted to tables1,000RecommendedPrevents stale data due to long-running metadata collection on the coordinator (full metadata collection).
Number of Parquet datasets promoted to tables5,000RecommendedPrevents stale data due to long-running metadata collection on executors (incremental metadata collection).
Minimum background refresh frequencyDuration of one refreshRecommendedBackground refresh frequency should be greater than the time it takes for a refresh to complete, otherwise refresh may run constantly.
Parquet footer16 MBSystemParquet footers are read on heap. If a Parquet file is mostly footer you should adjust ETL to create larger files.
Partitions per HDFS and Hive source300,000SystemHDFS and Hive sources only — to prevent excessive heap usage.
Promoted Delta or Iceberg tables5,000RecommendedPrevents stale data due to long-running metadata collection (incremental metadata collection).

Planner

ItemLimitLimit TypeAdditional Notes
Nodes in a query plan25,000SystemLarge plans can impact planning times, holding up incoming queries and impacting heap usage.
Number LimitsVaries by data typeSystemSee Data Types.
Varchar and Varbinary Limits32,000SystemSee Data Types.

Workload Management

ItemLimitLimit TypeAdditional Notes
Total active queries (queued and executing)1,000SystemIf more than 1,000 total active queries, consider adding another cluster with dedicated engines.