When you insert rows into a heap that has a nonclustered index and no clustered index, the insert goes to the last page of the table.
If the heap is partitioned, the insert goes to the last page on one of the partitions. Then, the nonclustered index is updated to include the new row.
If the table has a clustered index, it is used to find the location for the row. The clustered index is updated, if necessary, and each nonclustered index is updated to include the new row.
Figure 12-9 shows an insert into a heap table with a nonclustered index. The row is placed at the end of the table. A row containing the new key value and the row ID is also inserted into the leaf level of the nonclustered index.
Figure 12-9: An insert into a heap table with a nonclustered index
