This paper discusses the experience gained during the implementation of a prototype distributed database management system. The R* prototype supports transparent distribution of a relational database which is accessed using the SQL language for data definition and manipulation. While supporting autonomy among the participating sites, R* also emphasizes good performance for both local and multi-site operations. During the implementation of the R* prototype, we were guided by a few key objectives and encountered several interesting problems. This paper will discuss the guiding objectives of the R* effort, as well as several areas of the implementation which presented special difficulties or were simplified by design decisions. Copyright © 1987 by The Institute of Electrical and Electronics Engineers, Inc.