The multi-disk allocation problem is stated as follows: Allocate a given set of records among m ≥9T2 independently accessible disks in such a way that the average response time over all possible partial match queries is minimized, assuming that the retrieval of one record takes one unit of time. In a previous paper it has been shown that the problem is NP-hard. For the special case of a two-disk system, the present paper analyzes a heuristic method that first divides all records into equal-sized partitions and then allocates each partition randomly to one of the disks. For the expected response time, a formula is derived that can be used to show an interesting hierarchical performance ordering among all possible allocations by the above partition method.