“Using Experimental Measurements to Assess Dependable Adaptation Support Mechanisms for Timed Transactions”
in Workshop on Sharing Field Data and Experiment Measurements on Resilience of Distributed Computing Systems, with Proceedings of the 27th IEEE Symposium on Reliable Distributed Systems, Napoli, Italy, October 2008, Oct. 2008.
Abstract: Recently, we proposed a framework that allows to analyze stochastic data, namely end-to-end measurements of distributed interactions, and hence characterize the temporal probabilistic behavior of underlying communication or transactional services, to gather fundamental information for adaptation and dependability purposes. In particular, we argue that this framework can be used to provide support for some classes of timed transactions and it allows to satisfy important dependability objectives. This paper studies the framework in the scope of transactional environments, assessing the applicability of the proposed techniques for monitoring stochastic data, which is observed in these environments. We apply the framework using several sets of experimental field data that was collected with different kinds of transactions and in different load conditions. We compare the results with a baseline and conservative probabilistic approach and we show that the framework presents some benefits and may indeed be a useful tool to achieve more dependable timed transactions.
Research line(s): Timeliness and Adaptation in Dependable Systems (TADS)