Edge computing brings the service and utilities of cloud computing closer to end users. At the same time, the devices forming edge clusters are limited in featured computing resources, e.g., memory and storage. Running multi-component applications on edge clusters hence requires suitably selecting the nodes where to deploy the software stacks forming an application, so that they get actually hosted on cluster nodes featuring enough computing resources. In this paper, we illustrate a solution allowing to automatically determine feasible application deployments in edge clusters, i.e., automatically associating the software stacks forming an application to the nodes in a cluster that feature enough memory and storage for running them. Our solution is based on an existing graph transformation-based algorithm, with the main aim of demonstrating its potentials when applied to edge-related problems. Finally, we also illustrate an open-source prototype implementation of our solution.
Finding Feasible Application Deployments in Edge Clusters, with Limited Resources
Soldani J.
Primo
2021-01-01
Abstract
Edge computing brings the service and utilities of cloud computing closer to end users. At the same time, the devices forming edge clusters are limited in featured computing resources, e.g., memory and storage. Running multi-component applications on edge clusters hence requires suitably selecting the nodes where to deploy the software stacks forming an application, so that they get actually hosted on cluster nodes featuring enough computing resources. In this paper, we illustrate a solution allowing to automatically determine feasible application deployments in edge clusters, i.e., automatically associating the software stacks forming an application to the nodes in a cluster that feature enough memory and storage for running them. Our solution is based on an existing graph transformation-based algorithm, with the main aim of demonstrating its potentials when applied to edge-related problems. Finally, we also illustrate an open-source prototype implementation of our solution.I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.