This repository contains a reference implementation of the U-Sphere protocol. For more information about the protocol see the following paper:
- U-Sphere: Strengthening scalable flat-name routing for decentralized networks. http://dx.doi.org/10.1016/j.comnet.2015.07.006
In order to ease running of the distributed protocol testbed, a Docker
image has been provided in the Docker registry. It contains all the dependencies that are needed
to run the testbed. By default, the image will run the pf-b4 test, which uses a synthetic
128-node topology and performs some basic protocol tests (the StandardTests scenario).
After you have Docker and Docker Compose installed, you can simply start the testbed from the
top-level directory as follows (assuming unisphere repository is in ~/unisphere):
$ cd ~/unisphere
$ docker-compose up
This will pull all the required images (a MongoDB database is required) and run the above tests. The above instructions will work on any Linux distribution, which is able to run Docker containers. For more information about installing Docker and Docker Compose, see:
The scenario that is used by default will run for around 20 minutes.
The testbed is configured using a single settings file located at tools/settings.py. This
file contains the cluster configuration (by default a single machine is used for everything,
which is suitable only for emulating small topologies), topology configuration and scenario
configuration. For example the pf-b4 test run is defined under RUNS as follows:
dict(name="pf-b4", topology="basic_single", size=128, scenario="StandardTests"),This basically specifies that the basic_single topology generator should be used, which
is passed the size=128 parameter. It also specifies that the testbed should run the StandardTests
scenario.
Scenarios are defined as C++ classes in apps/testbed/scenarios.cpp and
they may call test cases which are defined in apps/testbed/tests.cpp.
After the test completes its execution, all results will be stored under a unique, randomly generated,
test run identifier (for example 5b9e0) in the output folder. The results contain whatever the scenario's
test cases output (usually CSV and GraphML files).