Towards Intelligent Cloud Diagnostics: Well Researched Software Marvel
A devoted group of researchers at North Carolina State University have painstakingly developed a novel software tool aimed at addressing performance disarrays in cloud computing systems. The tool functions to automatically classify and respond to potential network disruptions before they actually occur.
Cloud computing provides the freedom of creating numerous virtual machines provided to the end-users across a single computing platform – all that functions autonomously. Performance issues with such an approach are bound to occur. In case of a software glitch or a closely related hiccup, problems arising across a single effected virtual machine may end up bringing down the entire cloud down on its knees.
Determination of various contingencies across a system can be simplified by sensing and keeping a track of numerous machine related variables. The software does exactly that. By calculating the current network traffic, extent of memory consumption, CPU utilization, and several other parameters of data within a cloud computing infrastructure, the tool is able to estimate an effective measure of the overall system health. This renders the software flexible enough to formulate an adequate data-range characterization that can be safely considered as being normal. The processor usage, for instance, reflects the amount of computational power being required at any instant of time. The software outlines normal performance for every virtual machine in the cloud, and reports deviation of almost any sort. Based on the aforementioned information the tool predicts incongruities that might potentially affect the system’s capacity to provide service to users.
This particular approach is immensely beneficial in terms of associated benefits, including the all-important savings inherent with the alleviation of personnel training requirement. The software, being entirely autonomous depicts aberrant behavior on its own. In addition, the ability to predict anomalies is a feat that has never been achieved before. Not only that, upon sensing abnormal behavior in a virtual machine, it executes a pre-defined black box diagnostic test that determines which variables (memory usage, for instance) might be affected. The diagnostic data is then used to prompt the suitable prevention subroutine without making use of the user’s personal data in any form.
Helen Gu, co-author of the paper articulating this research marvel and an assistant professor at North Carolina State University explained: “If we can identify the initial deviation and launch an automatic response, we can not only prevent a major disturbance, but actually prevent the user from even experiencing any change in system performance.”
Most importantly, the software is not resource hungry (power in particular) and does not consume considerable amount of processor cycles to operate. It has the ability to fetch the preliminary data and classify normal behavior much quicker than the existing tactics. With CPU power consumption less than 1% of the total and a mere 16 megabytes of memory, the software is bound to pack a punch.
During the testing phase, the program recognized up to 98% of incongruities, which is the utmost as compared to existing approaches. It prompted a mere 1.7% of false alarms. Gu says: “And because the false alarms resulted in automatic responses, which are easily reversible, the cost of the false alarms is negligible.”
The software does sound like a real game-changer altogether. However, commercialization of the said research would eventually reveal the true benefits this tool has in store for the cloud computing industry – fingers crossed.
By Humayun Shahid