Pub/Sub Delivery Guarantees (PSDG) is a set of techniques that can be implemented in distributed publish/subscribe systems for message processing and delivery guarantees. Existing pub/sub systems that do not have these guarantees can only guarantee the delivery of messages to clients that are already known by all brokers. In the proposed algorithm, the brokers are connected to each other in an acyclic topology and messages are routed based on a reverse path forwarding scheme. In this implementation, the baseline without guarantees and the ACK based guarantee are implemented.

• Remy Duijsens
• Kristóf Oláh
• Tomás Herńadez Quintanilla

To run the project:

• Java 17
• Scala 2.13.8
• Maven

To run the simulation:

• Docker

To run the evaluation tool:

• Python 3 with ndjson, numpy and matplotlib installed

From the project root folder run mvn -f ./pom.xml clean compile package

From the project root folder run docker-compose build

From the project root folder run docker-compose up

In the .env file in the project root folder change the line to

DIRECTORY_SAVE=./experiments/runs/runs_[TYPE]/run[X]

where [TYPE] is either 'ack' or 'none', and [X] is the identifier of the run (1, ..., n).

After changing this setting, rebuild the docker environment and start the new run.

The evaluation tool folder is located at ./experiments/evaluation.

From this folder, run python3 run_evaluation.py. At least 2 runs are necessary to do the evaluation.

The results per run are stored in the ./experiments/evaluation/results folder. The accumulative results, namely the average and standard deviation over all runs, are located in ./experiments/evaluation/accumulative_results.

If the chosen guarantee type is changed, line 6 of the run_evaluation.py file has to be changed. Use option = "ack" for ack-based guarantee and option = "none" for no guarantee (baseline).

Copy the ack.json and none.json files from the ./experiments/evaluation/accumulative_results folder to the ./experiments/graph_scripts folder. From this folder, run python3 plot_chart.py. The resulting graph will be saved in the folder.

In the Client.scala class the number of simulations and the guarantee type can be changed that will be used in the simulations. The options are located at line 23 and 24.