Author: German Telmo Eizaguirre Suarez (URV, Tarragona) Version: 1.0 Date: 12-06-2019
Distributed System based on Serverless Computing in the IBM Cloud. This program uses the IBM Cloud Functions service through the IBM-PyWren middleware (http://cloudlab.urv.cat/josep/distributed_systems/p1-Sampe.pdf) and the Rabbit MQ service.
A finite number of functions are generated according to the number entered as argument. Each functions generates a random number that is transmitted to the other functions using a fanout Exchange. A coordinator ("master") guarantees mutual exclusion among processes and ensures all functions receive values in the same order. Finally, every serverless function returns the same list of values.
This project is implemented in Python 3.6
The main script is sdtask2.py.
For a correct execution of the system, a file cloud_config file with the RabbitMQ URL and some different setting must be set up. Follow the model at cloud_config_model.
For executing the program it is only necessary to call sdtask2.py as a python script. It receives the first argument as the number of nodes it has to create. The number of nodes must be wrapped between 1 and 17, if not it will be set to the default 5.
Example: >> python3 sdtask2.py 7
Different execution modes are allowed. They must be specified after the number of nodes. Posible modes are the following.
"-raw" results are simply collected, there are no extra logs.
"-sources" results are collected and the source node is showed along with each value from the list.
"-time" the total execution time of the program is showed at the end.
"-verbose" prints are done from the remote functions so that their workflow can be analysed from a remote log visualizer.
Example: >> python3 sdtask2.py 9 -sources
Example: >> python3 sdtask2.py 14 -time
Example: >> python3 sdtask2.py 12 -verbose
