Measuring consistency costs for distributed shared data

Distributed Shared Memory (DSM) systems typically sup- port one consistency protocol [3,5,6]. However, recent work [1,11,12,14,17] proposes the use of adaptive consistency based on a heuristical analy- sis of recent access patterns. Although heuristic-based approaches can significantly improve runtime, the access pattern alone does not nec- essarily define the most appropriate consistency protocol. The size of updates and other factors related to the computing environment, such as heavily loaded links, heavily loaded nodes, bursty traffic patterns, and network latency all affect performance. Multiple access patterns within the application also make it diffcult to select the most appropriate con- sistency protocol. This paper presents a measurement-based approach to the problem of selecting the most appropriate consistency protocol for the current application in the current runtime environment. We show that measurement-based analysis provides an accurate estimate of per- formance and therefore can be used to select the most appropriate con- sistency protocol, even in cases where hueristic-based approaches fail to select the appropriate protocol.