🌐 TrafPy 🚦

---

TrafPy is a Python package for the generation, management, and standardisation of network traffic.

Paper: Traffic generation for benchmarking data centre networks

Documentation: https://trafpy.readthedocs.io/

---

Projects which have used TrafPy

If you have used TrafPy in your project, please send an email with a link to [email protected] to have your work added to the below list!

• Y. Liu et al., 'A Hybrid Beam Steering Free-Space and Fiber Based Optical Data Center Network', Journal of Lightwave Technology, 2023
• C. W. F. Parsonson et al., 'A Vectorised Packing Algorithm for Efficient Generation of Custom Traffic Matrices', OFC'23: Optical Fiber Communications Conference and Exhibition, 2023
• C. W. F. Parsonson et al., 'Traffic Generation for Benchmarking Data Centre Networks', Optical Switching and Networking, 2022
• J. L. Benjamin et al., 'Traffic Tolerance of Nanosecond Scheduling on Optical Circuit Switched Data Centre Network', OFC'22: Optical Fiber Communications Conference and Exhibition, 2022
• J. L. Benjamin et al., 'Benchmarking Packet-Granular OCS Network Scheduling for Data Center Traffic Traces', Photonic Networks and Devices, 2021

---

Example

Generate simple flow-centric traffic for an arbitrary network you want to put under 10% traffic load with 16 leaf node end points which have communication link capacities of 1,250 information units per unit time. Save the traffic trace in a universally compatible file format so you can import it into any simulation, emulation, or experimentation test bed independently of TrafPy.

import trafpy.generator as tpg
import gzip, pickle

# init network
net = tpg.gen_arbitrary_network(num_eps=16, ep_capacity=1250)

# define flow characteristic distributions
node_dist = tpg.gen_uniform_node_dist(eps=net.graph['endpoints'])
interarrival_time_dist = {1.0: 1.0}
flow_size_dist = {1.0: 1.0}

# define network load config
network_load_config = {'network_rate_capacity': net.graph['max_nw_capacity'],
                       'ep_link_capacity': net.graph['ep_link_capacity'],
                       'target_load_fraction': 0.1}

# generate traffic demands
demand_data = tpg.create_demand_data(eps=net.graph['endpoints'],
                                     node_dist=node_dist,
                                     flow_size_dist=flow_size_dist,
                                     interarrival_time_dist=interarrival_time_dist,
                                     network_load_config=network_load_config,
                                     jensen_shannon_distance_threshold=0.9)

# save
Path('data/').mkdir(parents=True)
with gzip.open('data/demand_data.pickle', 'wb') as f:
    pickle.dump(demand_data, f)

See the tutorial and the examples directory for more information.

Install

Open Git Bash. Change the current working directory to the location where you want to clone this GitHub project, and run:

$ git clone https://github.com/cwfparsonson/trafpy

In the project's root directory, run:

$ python setup.py install

Then, still in the root directory, install the required packages with either pip:

$ pip install -r requirements/default.txt

or conda:

$ conda install --file requirements/default.txt

You should then be able to import TrafPy into your Python script from any directory on your machine

import trafpy

Citing this work

If you find this project or the associated paper useful, please cite our work:

@article{parsonson2022traffic,
    title = {Traffic generation for benchmarking data centre networks},
    journal = {Optical Switching and Networking},
    volume = {46},
    pages = {100695},
    year = {2022},
    issn = {1573-4277},
    doi = {https://doi.org/10.1016/j.osn.2022.100695},
    url = {https://www.sciencedirect.com/science/article/pii/S1573427722000315},
    author = {Christopher W.F. Parsonson and Joshua L. Benjamin and Georgios Zervas},
}

Contributing

File an issue here to report any bugs or new suggestions. Or, even better, for the repository on GitHub and create a pull request. If you want help making a pull request or are new to git, ask on the contributing issue you raise and/or see TrafPy's contributing guide.

License

TrafPy uses Apache License 2.0.