Xiao Liu PhD Thesis.pdf - Faculty of Information and Communication ...
Xiao Liu PhD Thesis.pdf - Faculty of Information and Communication ...
Xiao Liu PhD Thesis.pdf - Faculty of Information and Communication ...
Create successful ePaper yourself
Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.
work <strong>and</strong> problem analysis. Section 7.2 presents our novel adaptive temporal<br />
violation h<strong>and</strong>ling point selection strategy. Section 7.3 demonstrates the evaluation<br />
results. Finally Section 7.4 summarises this chapter.<br />
This chapter is partly based on our work presented in [56].<br />
7.1 Related Work <strong>and</strong> Problem Analysis<br />
7.1.1 Related Work<br />
Due to the dynamic nature <strong>of</strong> distributed computing infrastructures (e.g. p2p, grid<br />
<strong>and</strong> cloud) <strong>and</strong> the requirement <strong>of</strong> QoS, scientific workflow temporal verification<br />
has attracted increasing interests in recent years. Many efforts have been dedicated<br />
to different tasks involved in temporal verification. As mentioned earlier, a runtime<br />
checkpoint selection strategy aims at selecting activity points on the fly to conduct<br />
temporal verification so as to improve the efficiency <strong>of</strong> monitoring large scale<br />
scientific workflows <strong>and</strong> reduce the computation cost [21]. Meanwhile, to reduce the<br />
temporal violation h<strong>and</strong>ling cost, multiple-state based temporal verification is<br />
proposed to detect multiple fine-grained temporal violations so that different<br />
temporal violations (with different levels <strong>of</strong> time deficits) can be tackled by different<br />
temporal violation h<strong>and</strong>ling strategies [17].<br />
In recent years, many checkpoint selection strategies, from intuitive rule based to<br />
sophisticated model based, have been proposed. Most <strong>of</strong> them are introduced in<br />
Section 6.1.1 <strong>and</strong> hence omitted here. Among them, the state-<strong>of</strong>-the-art checkpoint<br />
selection strategies are those satisfy the requirements <strong>of</strong> necessity <strong>and</strong> sufficiency<br />
[21]. As mentioned before, the common philosophy in the studies on temporal<br />
verification is that a checkpoint is equal to a temporal violation h<strong>and</strong>ling point.<br />
Therefore, all these checkpoint selection strategies can be regarded as temporal<br />
violation h<strong>and</strong>ling point selection strategies. That is, the process <strong>of</strong> temporal<br />
violation h<strong>and</strong>ling point selection is actually overlooked in the existing temporal<br />
verification <strong>and</strong> temporal violation h<strong>and</strong>ling processes. Therefore, to the best <strong>of</strong> our<br />
knowledge, the issue <strong>of</strong> temporal violation h<strong>and</strong>ling point selection has being<br />
neglected <strong>and</strong> so far not been well addressed.<br />
105