Analysis and Comparison between Multiprocessing based and Middle ware based Parallel Remote administration Framework

Abstract: Remote administration is necessary in vast majority of enterprise infrastructure management. Installing software, applying patches or changing system configuration are all part of system administration job. Typically remote administration is done by accessing remote system through network by means of client-server protocol using locally available administration tools (remote terminal or remote desktop. Parallel remote administration frameworks are used in enterprise environment where similar administration tasks need to be performed on many systems. Different approaches are used to achieve remote administration on multiple machines in parallel. One of the method is using multiprocessing and other is using message oriented enterprise middle-ware. This paper gives the analysis of these two different approaches by comparing the performance, scalability, reliability of the framework in enterprise environment. The analysis is done against python „fabric‟ api for multiprocessing approach and „mcollective‟ orchestration framework for middleware based approach.

Keywords: remote administration, agent-based orchestration, fabric api, mcollective