Recents experiments have evaluated the ability of heterogeneous, hierarchical wireless sensor networks to scale to large node numbers and coverage areas. An exemplar is Project ExScal, where we designed and deployed a network of 1000+ sensor nodes and 200+ 802.11b backbone nodes for a 1.3km by 300m remote, open area. In this talk, we overview key issues in the scaling of wireless sensor network operations and applications, based on lessons derived from ExScal and other related experiments. We discuss at some length the impact of the network characteristics on a number of data management problems related to convergecast, broadcast, and in-network data flows. Finally, we identify challenges for future work in this area.