Mapping filter services on heterogeneous platforms

In this paper, we explore the problem of mapping filtering web services on largescale heterogeneous platforms. Two important optimization criteria should be considered in such a framework. The period, which is the inverse of the throughput, measures the rate at which data sets can enter the system. The latency measures the response time of the system in order to process one single data set entirely. Both criteria are antagonistic. For homogeneous platforms, the complexity of period minimization is already known [11]; we derive an algorithm to solve the latency minimization problem, and we provide a bi-criteria algorithm which minimizes latency without exceeding a prescribed value for the period. However, when adding heterogeneity to the platform, we prove that minimizing the period becomes NP-hard. The complexity for the latency is still open, even though we conjecture it might be NP-hard as well. We provide an integer linear program to solve both problems in the heterogeneous case. For the known NP-hard instance of the problem (period minimization, heterogeneous platforms), we design some efficient polynomial time heuristics and we assess their relative and absolute performance through a set of experiments. For small problem instances, the results are very close to the optimal solution returned by the integer linear program.