Because sequences are so pitifully slow.. Instead of paying the sequence cost with each insert, ArcGIS grabs one sequence value and allocates all the values in that range itself. If any values are left over, they are stuffed in a DBMS_PIPE for later reuse. Note that some sequences are set to 200 or even 400. If you restart the database, the pipe contents are lost, which could result in gaps. The Knowledge Base has documents which discuss both changing the interval and preserving cached ids.
- V
