• Neha V
• Anupama K C
• Dr R Nagaraja

Always failure jobs predictions with respect to existing and non-existing classification in service oriented architecture of cloud is always a big challenge. So we introduce a new 2 comparative approaches for predictions with the dataset which collected from Google clusters. The approach named as 2 level comparator (neural association LSTM, probabilistic regression).The goal of the work is to present an application failure prediction model that exactly predicts weather a task or job is successfully finished or failed. Considering the prediction on the traces of Google cluster data, there is a significant consumption of resource due to killed of failed jobs. For this we propose machine learning algorithm called LSTM (Long-Short-Term-Memory) network and by serializing the input we are comparing with PRM (probabilistic RegressionModel) for the accuracy. Prediction takes the priority and task attempts attributes of Google cluster data in order to predict the termination status (e.g., failed or finished). (keywords: long short term memory, serialization , regression prediction, failurejobs)

long short term memory, serialization , regression prediction, failurejobs

"The prediction dashboard on cluster data with comparative model for failed jobs", International Journal of Emerging Technologies and Innovative Research (www.jetir.org), ISSN:2349-5162, Vol.6, Issue 6, page no.368-372, June-2019, Available :http://www.jetir.org/papers/JETIR1906E04.pdf

"The prediction dashboard on cluster data with comparative model for failed jobs", International Journal of Emerging Technologies and Innovative Research (www.jetir.org | UGC and issn Approved), ISSN:2349-5162, Vol.6, Issue 6, page no. pp368-372, June-2019, Available at : http://www.jetir.org/papers/JETIR1906E04.pdf