On the Benefits of Profile-based Naming for Large Network Computing Systems

Naming and discovery are two critical issues of a widearea network computing system. The rising popularity of wide-area systems have resulted in the development of a variety of naming and discovery systems that are centered on a “descriptive” paradigm, where resources and services are described by a set of attribute-value tuples. This paper presents new evidence to suggest that descriptive naming systems of large-scale computing systems have special properties that can be exploited to construct very efficient and scalable implementations. We present the empirical evidence and perform simulations comparing a stock descriptive naming system with an enhanced naming system. Our simulation results indicate that the overhead caused by the naming system can be significantly reduced without any perceivable decline in performance.