The partitioning strategy used by a third party product results in a fairly sparse distribution of data.
i.e. of our ~900 partitions only ~300 have data. The way this works it will only become more sparse over time.
Is there any value in merging consecutive empty partitions (I can guarantee there will be no inserts into them) or is this a waste of time?
Sql-server – Are there any issues with sparsely populated partitioned tables
partitioningsql server
Related Question
- Sql-server – How to do data migration between different schemas
- SQL Server 2008 Partitioning – How to Reduce Table Partition Timing
- Local indexes vs Global indexes for partitioned tables in Oracle
- Sql-server – Optionally exclude certain partitions using OLA’s scripts
- SQL Server Best Practice – Keeping Empty Partitions at Both Ends of Table
- PostgreSQL 11 – How to Partition 1000 Tenants
Best Answer
Each partition is essentially a separate table, so there is some cost involved in query optimization, statistics, and metadata management. So if the third-party product supports you removing partition boundary points they created, I would probably do it.
See
Partitioned Tables and Indexes - Performance Guidelines