Abstract
GUIDs are globally unique identifiers that can be generated independently. They play an important role in many distributed systems. However, compared to most local identifiers, they are relatively large (16 bytes). We discuss techniques that can make the use of GUIDs more efficient, so that they become more viable as a ubiquitous solution for large distributed databases and similar applications relying on artificial identifiers. The proposed techniques include canonical GUIDs for primitive values and a compression scheme for timestamp-based GUIDs. The techniques have been implemented and can be used in different programming languages. Furthermore, general aspects and potential problems of GUIDs as defined in the RFC 4122 are discussed.