Currently, only HG, UNIQUE HG, UNIQUE CONSTRAINT, and PRIMARY KEY indexes support multiple columns in index creation, but multi-column indexes are also useful for GROUP BY and ORDER BY statements.
From a statistics point of view, multi-column indexes provide enough information in multi-column table joins to let the optimizer know the exact statistics of the join and whether or not it is a many-to-many or one-to-many join. The optimizer is also smart enough to use the statistics for optimization, but use individual HG/LF indexes for the actual work. The optimizer costs out all join and sort scenarios and decides which index(es) is best for that operation. The statistics help it get to that point.
Try to guarantee that inserts will happen at the end of the index
Place generally incrementing data, like a transaction date or batch number (sequential data), at the beginning of the index list. Something that will try to guarantee a sequential key
See the previous section the HG index loads.