The bottleneck is probably the number of processors and/or the number of containers per tablespace.
I usually set it equal to the number of containers that reside on separate physical disks per tablespace. For RAID disks, it would be the number separate arrays in a container. If you only have one container in a tablespace, you will get improved performance from the striped EMC SAN, but I don't think DB2 itself will initiate parallel I/O in that situation (for a single query).
It definitely helps if you have at least one processor for each I/O server, although this may not be absolutely necessary. If you have enough processors, then you could set it equal to 2x, 3x, etc the number of containers in a tablespace.