A customer wants to expand his cascaded B-series based SAN but is concerned about latencies within the SAN backend. You recommend changing the topology to a full mesh.

Which switch feature will ensure that the latencies in the backend are at the lowest possible level?