TopoHub: A repository of real-world topologies for networking research

Repository of reference Gabriel graph, Internet Topology Zoo, SNDlib, CAIDA and synthetic backbone topologies for networking research. Synthetic Gabriel graphs very well represent properties of WAN / long-haul optical transport networks. The repository offers 3600 Gabriel graph topologies with linearly increasing sizes ranging from 5 to 500 nodes, generated in a reproducible manner. Topologies are available in the code repository and can be previewed and downloaded through a web interface, which allows visualization of individual topologies and exploration of their network properties. An important additional feature is the visualization of pre-computed link loads in the network using the Equal-Cost Multipath (ECMP) shortest path routing algorithm under different traffic demand models.

Web interface is available at: https://www.topohub.org

The package also includes a module that can be imported into the popular network emulator Mininet, enabling automatic usage of the topologies from the repository.

Please cite the following paper if you use of TopoHub in your research:

@article{topohub,
    title = {TopoHub: A repository of reference Gabriel graph and real-world topologies for networking research},
    journal = {SoftwareX},
    volume = {24},
    pages = {101540},
    year = {2023},
    issn = {2352-7110},
    doi = {10.1016/j.softx.2023.101540},
    author = {Piotr Jurkiewicz}
}

The Python package can be installed from Python Package Index (PyPI) using the following command:

pip install topohub

Then you can obtain topologies stored in the repository using the topohub.get() method and create NetworkX graph objects basing on them:

import networkx as nx
import topohub

# Obtain topology dicts from JSON files stored in the repository
topo = topohub.get('gabriel/25/0')
topo = topohub.get('backbone/africa')
topo = topohub.get('topozoo/Abilene')
topo = topohub.get('sndlib/polska')

# Create NetworkX graph from node-link dict
g = nx.node_link_graph(topo, edges='edges')

# Access graph parameters
print(g.graph['name'])
print(g.graph['demands'])
print(g.graph['stats']['avg_degree'])

# By default, nodes are referred by their integer IDs
# You can obtain node names through the `name` attribute
print(g.nodes[0]['name'])
print(g.nodes[10]['name'])

# Obtain link length in kilometers between node 0 and 10
print(g.edges[0, 10]['dist'])
# Obtain percentage utilization of the link between node 0 and 10 under ECMP routing in the forward direction
print(g.edges[0, 10]['ecmp_fwd']['uni'])

# You can also load a topology using node names instead of integer IDs as node identifiers
# (this will not work for 'backbone' and 'caida' topologies which have unnamed or duplicated name nodes)
topo = topohub.get('sndlib/polska', use_names=True)
g = nx.node_link_graph(topo, edges='edges')

print(g.graph['demands'])
print(g.edges['Gdansk', 'Warsaw']['dist'])
print(g.edges['Gdansk', 'Warsaw']['ecmp_fwd']['uni'])

For usage in Mininet, you can use a helper which automatically creates Mininet Topo classes for selected topologies:

import mininet.net
import topohub.mininet

# Obtain Mininet Topo classes for topologies stored in the repository
topo_cls = topohub.mininet.TOPO_CLS['gabriel/25/0']
topo_cls = topohub.mininet.TOPO_CLS['backbone/africa']

# Alternatively, you can also load a topology using node names instead integer IDs as node identifiers
# (this will not work for 'backbone' category topologies which have unnamed or duplicated name nodes)
topo_cls = topohub.mininet.TOPO_NAMED_CLS['topozoo/Abilene']
topo_cls = topohub.mininet.TOPO_NAMED_CLS['sndlib/polska']

# Initialize Mininet Topo object
topo = topo_cls()
# Create Mininet Network using the selected topology
net = mininet.net.Mininet(topo=topo)
# Start the network and Mininet shell
net.interact()

Detailed documentation, including API reference and Mininet usage example, is available at: https://topohub.readthedocs.io