• Kalyani Sanjay Vispute
• Dr. D.S. Waghole

Scheduling and resource provisioning are two tactics that require the ability to predict how workflow processes will respond to changes in input data. However, it is difficult to generate such approximations in the cloud. This research proposes a novel two-stage machine learning approach for predicting the execution time of cloud workflow tasks for various input data. To provide high accuracy forecasts, our method relies on two stages of prediction and parameters providing runtime information. Our method outperforms current prediction methods, as demonstrated by empirical results for four real-world workflow applications and a variety of commercial cloud providers. Comparatively, prior methods produced estimation errors of greater than 20% (and often greater than 50%) in more than 75% of the workflow tasks that were looked at. In our experiments, our technique yields 1.6% estimate errors in the best-case scenario and 12.2% estimation errors in the worst situation. We also show that the models predicted by our method for a certain cloud can be reliably and readily transferred to other clouds with a few numbers of executions.

Performance Prediction, Workflow Execution, Task Distribution

"Determining the Duration of Cloud Workflow Task Execution using a Two-Step Machine Learning Approach", International Journal of Emerging Technologies and Innovative Research (www.jetir.org), ISSN:2349-5162, Vol.11, Issue 4, page no.g509-g516, April-2024, Available :http://www.jetir.org/papers/JETIR2404668.pdf

"Determining the Duration of Cloud Workflow Task Execution using a Two-Step Machine Learning Approach", International Journal of Emerging Technologies and Innovative Research (www.jetir.org | UGC and issn Approved), ISSN:2349-5162, Vol.11, Issue 4, page no. ppg509-g516, April-2024, Available at : http://www.jetir.org/papers/JETIR2404668.pdf