Abstract:

A large-scale cloud data centre must deliver high service dependability and availability while minimising failure incidence. However, modern large-scale cloud data centers continue to have significant failure rates owing to a variety of factors, including hardware and software faults, which often result in task and job failures. Such failures may substantially degrade the dependability of cloud services while also using a large amount of resources to restore the service. To reduce unexpected waste, it is critical to forecast task or job failures with high accuracy before they occur. Many machine learning and deep learning-based approaches for task or job failure prediction have been presented, which include examining previous system message logs and detecting the link between the data and the failures. In this research, we present a failure prediction technique based on multi-layer Bidirectional Long Short Term Memory (Bi-LSTM) to detect task and job failures in the cloud, in order to enhance the failure prediction accuracy of prior machine learning and deep learning-based approaches. The purpose of the Bi-LSTM prediction algorithm is to anticipate whether tasks and jobs will be completed or unsuccessful. Our approach beats existing stateof-the-art prediction algorithms in trace-driven tests, with 93% accuracy for task failure and 87% accuracy for job failures, respectively

Description:

Cloud datacenters and deep learning.