Design and evaluation of redistribution strategies for wide-area commodity distribution

The proliferation of e-commerce has enabled a new set of applications that allow globally distributed purchasing of commodities such as books, CDs, travel tickets, etc., over the Internet. These commodities can be represented online by tokens, which can be distributed among servers to enhance the performance and availability of such applications. There are two fundamental approaches for distributing such tokens-partitioning and replication. Partitioning-based approaches eliminate the need for tight quorum synchronization required by replication-based approaches. The effectiveness of partitioning, however, relies on token redistribution techniques that allow dynamic migration of tokens to where they are needed. We propose pair-wise token redistribution strategies to support applications that involve wide-area commodity distribution. Using a detailed simulation model and real Internet message traces, we investigate the performance of our redistribution strategies and a previously proposed replication based scheme. Our results reveal that, for the types of applications and environment we address, partitioning-based approaches perform superior primarily due to their ability to provide higher server autonomy.