SAP HANA – Replicated table with LOB columns is much bigger than in source system

In last time I observed in HANA system worrying growth a table replicated by SDI which size is now above 100GB. The table at source system is only 2.8GB. It’s worth noting:

• DB source is MSSQL;
• three of the columns are NCLOB datatype;
• We reduced the size of the table by implementing note: 2950474 but the size is growing again.

Solution

Useful information You can receive from below reports.

Check the output of below query :

CALL CHECK_TABLE_CONSISTENCY('CHECK_HYBRID_LOB_OVERHEAD', '', '');

Also run the below reports from the attached SQL in the Note 1969700 :

1. Overview of disk LOB files;
2. HANA_Disks_Overview*;
3. HANA_Disks_Data_Overview*;
4. HANA_Tables_DiskSize*;
5. HANA_Tables_LargestTables*;
6. HANA_Tables_ColumnStore_Overview*;
7. HANA_Tables_TopGrowingTables_Size_History;

As I earlier mentioned we reduced the size of the table by implementing note: 2950474 – was executed the repair job: REPAIR_HYBRID_LOB_OVERHEAD. Unfortunately the size is grew again.

Finally it turned out this is a known issue that table consistency check methods CHECK_HYBRID_LOB_OVERHEAD and REPAIR_HYBRID_LOB_OVERHEAD do not correctly report and repair disk level fragmentation of packed LOB containers. This causes affected packed LOB columns to consume more space on disk than necessary.

Affected Releases:

• SAP HANA 2:
  • Revisions <= 048.03 (SPS04)
  • Revisions <= 054.00 (SPS05)

Solution:

Belongs apply one of the following SAP HANA Revisions:

• SAP HANA 2:
  • Revisions >= 048.04 (SPS04)
  • Revisions >= 055.00 (SPS05)
  • or higher

Refer more details in the attached Note 3012162 – Table Consistency Check CHECK_HYBRID_LOB_OVERHEAD and REPAIR_HYBRID_LOB_OVERHEAD Fail to Check all Packed LOB Containers.