RESEARCH THEME 03 / DOCTORAL WORK

End-to-end delay analysis.

Modelling the time taken for data to travel through a sequence of stream-processing operations.

The approach

A continuous query does more than compute a value: it also determines when that value becomes available. Understanding total delay requires a model of how data moves between operators and how their executions are scheduled.

This work uses causal relationships between data tokens and periodic execution schedules to calculate end-to-end delays. The paper presents the model, an algorithm, and an experimental evaluation.

Related publications

  1. 2015
    JOURNAL ARTICLE

    Computing end-to-end delays in stream query processing

    Vasvi Kakkad, Andrew E. Santosa, Alan Fekete, Bernhard Scholz

    Science of Computer Programming, 105, pp. 124–144

    A model and algorithm for calculating the time taken for data to move through a stream query. The work uses causal relationships between data tokens and periodic execution schedules, with experiments evaluating the approach.