Abstract:
Tenants want automatic MapReduce resource estimation tools to move cloud computing from pay-per-request to pay-per-use. These tools require accurate workload, resource, and completion time quantification. Prediction models exist.
None of these models account for virtual machines’ (VMs) performance variance during job execution, resulting in underestimating resources and exceeding deadlines. We suggest a multi-view deep learning model to capture real-time performance variance and automatically scale out the cloud cluster.
MarVeLScaler, a prototype system with two useful modules, Scale Estimator and Scale Controller, is implemented. Scale Estimator estimates the MapReduce cluster size for a specific workload and deadline.
Scale Controller adjusts cluster scale based on real-time running status to ensure job completion. We test Hadoop-based MarVeLScaler in Alibaba Cloud. Experiments show that MarVeLScaler can predict initial cluster size with 98.4% accuracy, save 30.8 percent, and perform similarly to state-of-the-art methods.
Note: Please discuss with our team before submitting this abstract to the college. This Abstract or Synopsis varies based on student project requirements.
Did you like this final year project?
To download this project Code with thesis report and project training... Click Here
