A rule-fitting solver, which takes an OpenFlow 1.3 ruleset and transforms it to fit a new constrained fixed-function pipeline.
The main executable is ofsolver and is installed to
/usr/bin in your python environment.
To use the recommended settings run:
ofsolver --progress ruleset.pickle ttp.json
Where ttp.json is a Table Type Pattern encoded as JSON and ruleset.pickle is in the format supported by ofequivalence. You can collect a ruleset in this format from a switch using this script: collect_state.py.
There are some sample rulesets and patterns in tests/, e.g.:
ofsolver --progress tests/rulesets/sample_rules.pickle tests/ttps/test_pipeline2.json
For more advanced settings check --help:
usage: ofsolver [-h] [-t] [-d] [-s] [-a {Search,SAT,SSAT,SingleSAT,Genetic}]
[-p] [--progress] [-l {CRITICAL,ERROR,WARNING,INFO,DEBUG}]
[--ttp-log-level {CRITICAL,ERROR,WARNING,INFO,DEBUG}] [-C]
[-f | -b] [-i ITERATIONS] [-M] [-R] [-G]
[-o {NO_CONFLICT,NO_HIT,NO_MISS,NO_PLACEMENT,NO_PLACEMENT_CONFLICT,NO_SAME_TABLE}]
capture ttp
Transforms an OpenFlow ruleset to fit a fixed-function pipeline
positional arguments:
capture A file of a pickled set of flow rules, either a full
state capture or list of flow stats
ttp A JSON Table Type Pattern description
optional arguments:
-h, --help show this help message and exit
-t, --time Collect timing information
-d, --dummy Use dummy multiprocessing i.e. threading
-s, --single Use the single threaded solver
-a {Search,SAT,SSAT,SingleSAT,Genetic}, --algorithm {Search,SAT,SSAT,SingleSAT,Genetic}
The algorithm to use
-p, --print-failure Prints additional information about a failure
--progress Track progress throughout the solver, adds progress
bars to loops.
-l {CRITICAL,ERROR,WARNING,INFO,DEBUG}, --log-level {CRITICAL,ERROR,WARNING,INFO,DEBUG}
The solvers logging level
--ttp-log-level {CRITICAL,ERROR,WARNING,INFO,DEBUG}
The logging level used when loading the TTP
-C, --no-compression Disable compression of the input ruleset.
-f, --full Collect all valid results
-b, --best Check all results, only save the best and worst
-i ITERATIONS, --iterations ITERATIONS
Limit the iterations of the solver
-M, --no-merge Disable merge placements
-R, --no-reactioning Disable reactioning. Reactioning tries to replace a
placement with a placement from another rule with the
same match but different actions. See
ofsolver.solver.Solver.reactioning
-G, --no-generalisation
Disable generalisation. Generalisation creates rules
with less specific matches. See
ofsolver.solver.valid_generalisation.
-o {NO_CONFLICT,NO_HIT,NO_MISS,NO_PLACEMENT,NO_PLACEMENT_CONFLICT,NO_SAME_TABLE}, --option {NO_CONFLICT,NO_HIT,NO_MISS,NO_PLACEMENT,NO_PLACEMENT_CONFLICT,NO_SAME_TABLE}
Additional solver specific options
Install the C library requirements for gmpy2 as required by ofequivalence. For Debian based distributions run:
apt install libgmp-dev libmpfr-dev libmpc-dev
Install the python requirements. In the root directory (containing this readme) run:
pip install -r requirements.txt
Download the required minisat-zmq and muser2 binaries, these are placed locally in the ofsolver directory. Alternativelly see the documentation of those projects and install the binaries to the system path.
./setup.py download
Install the ofsolver tool and library (use the pip --user option to install for only the local user):
pip install .
Note: tests require minisat to be installed to the system path (on debian based systems apt install minisat2).
Unittest is used to run tests.
In the root directory, the following command will run the tests locally:
./setup.py test
Or alternatively:
python -m unittest discover -v
The code is licensed under the Apache License Version 2.0, see the included LICENSE file.
React
Typescript
Django
Laravel
Facebook
Microsoft
Google
Alibaba
D3
Tencent