TEAVAR is a traffic engineering scheme that balances utilization and availability to provide bandwidth guarantees at a desired SLA. This codebase also includes a simulation framework that was used to test TEAVAR, as well as implementations of other traffice engineering algorithms in Julia.
Quick start:
-
Install Julia
-
Install open a julia shell using:
julia -
Install JuMP version v0.18.5 using the Julia REPL:
julia> ] (v1.0) pkg>add JuMP@0.18.5 -
Include certain files using:
include("./main.jl")
main.jl is the main entrypoint but you can also include specific files invidually to run specific experiments
Note: Most TE algorithms in this codebase are implemented as Linear Programs and require the Gurobi solver.
Using Pre-built binaries:
Instead of installing Julia and using the Julia REPL to run different files, we have compiled the codebase into binaries available on Mac and Linux. To install these go to our homepage or follow the links above.
-
Install the binary for your operating system
-
Unzip the file
tar -xvf linuxbuild.tar.gz
- Run the file teavar. Add -help or -h for more information on accepted parameters
./teavar -help
Data: All data used for TEAVAR evaluations is in the data folder. Each folder contains a topology. A topology must include 3 files.
- nodes.txt A list of string nodenames
- topology.txt A list of rows containing edges with a source, destination, capacity, and probability of failure.
- demand.txt A set of demands in matrix form or list form. (See data/Custom/demand.txt for list form)
Results: Results from experiments are saved in data/raw. Each experiment has its own folder in data/raw that contains a counter and saves each experiment run results in a folder with the counter number. Results used in the paper have been moved from data/raw to gnuplot/data.
Plots: Data for plots used in TEAVAR can be found in the gnuplot/data folder. All plotting code is also in gnuplot folder (gnuplot/*.gp) and each plotting file outputs plots to gnuplot/plots.
Gnuplot can be installed on OSX using homebrew (brew install gnuplot) or linux using apt (apt get gnuplot). More information on gnuplot can be found on the gnuplot website.
TE Schemes: All TE schemes (mostly case as linear programs) can be found by name in the Algorithms folder.
See the list contributors.