STATSFEED: An Extensible Statistics Collection Framework for Virtualized Environment

Abstract

-The main aim of this project is focused on collecting the system logs (CPU usage, memory usage, resource utilization, etc.) which is designed to be used on scalable virtualized environment. For collecting the logs from the hosts and the virtual machines, a performance manager program is used which is written in java using VMware infrastructure API and these hosts are placed on the ESXi hypervisor which is used as server, it is of the type bare metal/type1 hypervisor. This is connected to the vCenter and it gathers the logs from the hosts and the virtual machines. The logs from every VM and host are written into discrete text records appended with the name of those machines. The system is intended to accumulate the insights uncertainly, until it is disturbed on manually. The statistical factors that are gathered which identifies the workloads on the machines are stored in the scalable database called MsSQL(Microsoft SQL Server) which contains both historical as well as transactional data, and then these real time statistics is consolidated and displayed in the form of graphs visually by using the visualization tool like solarwinds etc.