A decentralized resource management technical framework aiming to deploy applications at the edge of the network, guaranteeing adherence to (i) defined latency Service Level Agreements (SLAs) and (ii) resource preferences of participating edge nodes.
Download the framework and install all requirements from the requirements.txt file. The project was developed in Python 2.7.
First, the user must provide two different input JSON files:
- application model file, where the application's requirements and communication path is described, stored in the app folder.
- target edge architecture file, stored in the edgeArchitecture folder. Here, provide the IP addresses of the edge nodes where the API resides or use localhost if the nodes run locally on the same machine.
In our framework there are two important entities, i.e., the coordinator node and the coolaborator nodes:
- dispatcher.py: represents the functionality of the coordinator node.
- edgenode_api_main.py: represents the functionality of the collaborator nodes.
More details can be found in our technical paper.
First, the edgenode_api_main.py must run on the collaborators. To do so, we can use the following command:
python edgenode_api_main.py <port_name>By default, if <port_name> is omitted, the port will be set to 5000.
Once all collaborators are active and the two imput files are prepared, we can execute dispatcher.py, using the following command:
python dispatcher.py -a "antivirusApp" -e "edgeNodes" -sIn the command above we tell the dispatcher that the application model file name is antivirusApp and the target edge architecture is described in edgeNodes JSON file. Finally, to see the deployment strategy on the stdout, we add the -s argument.
For a complete list of arguments that can be parsed as input to dispatcher.py please run:
python dispatcher.py -h