A Page Split occurs in B-trees (the underlying structure of most database indexes) when a new record is inserted into a full data page. To make room, the database engine splits the page into two, redistributes the records, and updates the index pointers.
- Breaks Physical Order – The new page may be allocated in a different location, disrupting the sequential structure of the data.
- Extra Redistribution Costs – Each split requires modifying adjacent pages and potentially updating parent nodes, increasing system load.
- Decreased Read Efficiency – Fragmentation forces the database to perform additional I/O operations when reading sequential data.