Explore chapters and articles related to this topic
Configuration and Usage of Open-Source Protocol
Published in Ivan Cibrario Bertolotti, Tingting Hu, Embedded Software Development, 2017
Ivan Cibrario Bertolotti, Tingting Hu
Both strategies are useful in principle because they represent different trade-offs between flexibility, robustness, and efficiency. For instance, the allocation of a fixed-size data structure from a dedicated pool holding memory blocks of exactly the right size can be done in constant time, which is much more efficient than leveraging a general-purpose (but complex) memory allocator, able to manage blocks of any size.
Multi-objective auto-scaling scheduling for micro-service workflows in hybrid clouds
Published in Enterprise Information Systems, 2023
Shijia Wang, Xuan Liu, Ming Gao, Mingxia Chen, Kai Leung Yung, Shancheng Jiang
Part 2: The cloud centre Controller can perform three different operations: lease a new VM instance and deploy the container; terminate a VM instance without a container deployed, i.e. release all resources of the instance; lease a previously used VM instance and perform a new round of container deployment. After the Controller receives the scheduling request, the Docker Allocator in it performs further scheduling tasks, selecting the corresponding Docker container for each micro-service workflow. The Cloud Manager will select a Cloud for each container and provide available VMs. We give priority to specifying previously deployed VMs, thus saving time in starting the image when the Docker container is first deployed. Finally, the Cloud Monitor Controller monitors the running and released Docker containers on the VMs in real time, and provides information such as the list of running Docker containers on each VM and their running status to the Docker container allocator to complete the next round of scheduling.
Control allocation-based adaptive control for greenhouse climate
Published in International Journal of Systems Science, 2018
Yuanping Su, Lihong Xu, Erik D. Goodman
From Figure 2, we can observe that the solutions in region 1 basically favour objective J1(u, x, w), while the solutions in region 2 favour objective J2(u, x, w). If , we must choose a solution from region 2 to ensure the convergence of the system; otherwise, an appropriate preference in region 1 is selected to minimise objective J1(u, x, w). For this purpose, the two relative change rates for the two objectives are introduced in order to allow evaluating the preference performance of the solutions in the optimal population; these are defined, respectively, as where , are the solutions with which the objectives J1(u, x, w) and J2(u, x, w) reach their maximum, respectively, and ui represents the ith solution in the optimal population. Next, the mean rank of the tracking errors is calculated by Finally, the decision strategy can be summarised as follows: if else if else where , σJ1max , σJ2max are the maximum relative change rates in the optimal Pareto solution set, respectively, Jmin 2 is the minimum of objective J2(u, x, w) in the optimal Pareto set, and upre represents the output of the allocator.