Dimensionality Reduction Algorithms for Scalable Monitoring of Virtual Machines

In the past the statistical technique of Principal Component Analysis has been used for finding relevant information to be used for monitoring virtual machine clustering. We intend to use other dimensionality reduction algorithms for the same namely Latent Dirchlet Allocation,Autoencoders, and backward elimination to find a better alternative for the same. We propose a method wherein we try to incorporate the data obtained from these executions and then reduce its dimensionality or in Layman terms we try to minimize the monitor-able data to a minimum so as to efficiently monitor data. This will not only reduce the time taken for cloud monitoring but also help in scaling virtual machine clusters as the data obtained for monitoring is directly proportional to number of clusters and the application, hence if we are able to minimize the data to be monitored then we can increase the amount of clusters and still be able to maintain a proper monitor-able environment with lesser variables to be concerned about.