DB2 9 for z/OS offers relief in this area by allowing the insert of the sequential column through a new RANDOM option on both the CREATE INDEX and ALTER INDEX statements. When RANDOM is used on the column in an index, DB2 stores the key values in random locations within the index tree. This reduces the chance of ever-increasing sequential inserts being on the same page, which decreases the chance of having lock contention.
Randomly ordered indexes can be used with equality lookups on a specific column. In addition, randomly ordered key columns can be used in non-matching index scans, while index-only access can be used in random key columns.
Here's an example of a CREATE INDEX using the RANDOM clause:
CREATE INDEX creator.table_name_idx1
RANDOM cannot be specified on a key column that is varying in length with the NOT PADDED option. If you've taken advantage of RANDOM key columns in your application, please take a moment and tell us of your experience in Comments.