Location-based service (LBS) systems have recently become a subject of active research and development. A key issue in such systems is efficient management of current and past location information of moving items. In the past two years, we have been establishing a new LBS system architecture, named the GALIS (Gracefully Aging Location Information System), which is based on the new approach of using a cluster of database server nodes and effecting efficient distributed and parallel computing in both real-time update of location records of moving items and processing of location-related queries. Each server node is dedicated to keeping records relevant to a different geographical zone and a different temporal zone. Each subsystem in GALIS is structured as a TMO network for the sake of easy analysis and expansion. A prototype implementation of GALIS has been realized only partially, and its full implementation is under way. In this paper, major features of GALIS are presented together with discussions on trade-off considerations which led to the adoption of those features. Issues remaining for further research are also briefly discussed.