Each object in the shared data space must have a unique name. A global name space is necessary for caches to correctly associate updates with their respective cached objects. Unique object names are generated by servers when an update is received for an unnamed object. When a server detects an update for a new object, it tags the object with the unique name ``server:port:count.'' Since no two servers can be listening to the same port on the same machine, the prefix distinguishes the server stamping the object. The suffix is an integer that enumerates the objects stamped by the same server. Together they form a unique name for every object in the system.