|**trim_reason_desc**|**nvarchar(60)**|Description of *trim_reason*.<br /><br /> 0 - UNKNOWN_UPGRADED_FROM_PREVIOUS_VERSION: Occurred when upgrading from the previous version of [!INCLUDE[ssNoVersion](../../includes/ssnoversion-md.md)].<br /><br /> 1 - NO_TRIM: The row group was not trimmed. The row group was compressed with the maximum of 1,048,576 rows. The number of rows could be less if a subset of rows was deleted after delta rowgroup was closed<br /><br /> 2 - BULKLOAD: The bulk-load batch size limited the number of rows.<br /><br /> 3 - REORG: Forced compression as part of REORG command.<br /><br /> 4 - DICTIONARY_SIZE: Dictionary size grew too large to compress all of the rows together.<br /><br /> 5 - MEMORY_LIMITATION: Not enough available memory to compress all the rows together.<br /><br /> 6 - RESIDUAL_ROW_GROUP: Closed as part of last row group with rows < 1 million during index build operation. Note: A partition build with multiple cores can result in more than one trim of this type.<br /><br /> 7 - STATS_MISMATCH: Only for columnstore on in-memory table. If stats incorrectly indicated >= 1 million qualified rows in the tail but we found fewer, the compressed rowgroup will have < 1 million rows<br /><br /> 8 - SPILLOVER: Only for columnstore on in-memory table. If tail has > 1 million qualified rows, the last batch remaining rows are compressed if the count is between 100k and 1 million<br /><br /> 9 - AUTO_MERGE: A Tuple Mover merge operation running in the background consolidated one or more rowgroups into this rowgroup.|
0 commit comments