What is it about?
In this article, we introduce a new algorithm KaliGreen that can maneuver the microservices within a network of devices in order to maximize the run-time of a microservice-based application; moreover, KaliGreen allows a 54% increase in the average run-time of an application by shifting microservices from 6 devices (as example) with low battery or inefficient processing ratios to devices in better conditions. To achieve this, KaliGreen utilizes KaliMucho middleware, which is able manipulate microservices in run-time. This algorithm provides a plausible solution to maximizing energy consumption within a network of devices.
Featured Image
Why is it important?
In this article, we have presented KaliGreen.This is a middleware with a non-centralized conscious scheduling algorithm that monitors the energy levels of user's devices with the objective of saving energy and battery in order to allow applications to run as long as possible. In the continuous executions in our simulator, we saw with satisfaction that the devices that depend on the battery in conditions and overload of CPU or RAM or network were freed of load in an intelligent way, prioritizing devices with dangerous battery conditions. Thus, an application with a limited duration of time due to battery conditions, for example of 6 devices, can be redistributed and continue to run for 54% more of time on the other user devices if KaliGreen is launched.
Perspectives
- One of the objectives of KaliGreen is not to affect the user experience when operating. In this article, it is considered that no matter how intelligent the heuristics are about knowing the behavior pattern of a person, there will always be dynamic situations in which he will be affected in efficiency and time. We consider future work, studying the best methods to move microservices without affecting the user experience. - In order to select the most expensive microservice in energy, it is necessary to know its consumption ratio in terms of RAM, CPU and network. However, it is also necessary to find the proportional indexes of the consumption of each component (In the Section \ref{subsection:ANS}, rather than setting these values manually). A future work, is to find those coefficients dynamically depending on the hardware and operating system.
Hernan Humberto ALVAREZ VALERA
Read the Original
This page is a summary of: KaliGreen: A distributed Scheduler for Energy Saving, Procedia Computer Science, January 2018, Elsevier,
DOI: 10.1016/j.procs.2018.10.172.
You can read the full text:
Contributors
The following have contributed to this page
