Please see https://github.com/initc3/HoneyBadgerBFT-Python/ for the more up-to-date Python implementation, and https://github.com/initc3/HoneyBadgerBFT/ for an umbrella project about the HoneyBadgerBFT protocol.
The Honey Badger of BFT Protocols
Most fault tolerant protocols (including RAFT, PBFT, Zyzzyva, Q/U) don't guarantee good performance when there are Byzantine faults. Even the so-called "robust" BFT protocols (like UpRight, RBFT, Prime, Spinning, and Stellar) have various hard-coded timeout parameters, and can only guarantee performance when the network behaves approximately as expected - hence they are best suited to well-controlled settings like corporate data centers.
HoneyBadgerBFT is fault tolerance for the wild wild wide-area-network. HoneyBadger nodes can even stay hidden behind anonymizing relays like Tor, and the purely-asynchronous protocol will make progress at whatever rate the network supports.
This is released under the CRAPL academic license. See ./CRAPL-LICENSE.txt Other licenses may be issued at the authors' discretion.
Build the docker image first.
cd docker
docker build -t honeybadgerbft .
Then for example you want to run an instance with N=8, t=2 and B=16:
docker run -e N="8" -e t="2" -e B="16" -it honeybadgerbft
Working directory is usually the parent directory of HoneyBadgerBFT. All the bold vars are experiment parameters:
- N means the total number of parties;
- t means the tolerance, usually N/4 in our experiments;
- B means the maximum number of transactions committed in a block (by default N log N). And therefore each party proposes B/N transactions.
pbc
wget https://crypto.stanford.edu/pbc/files/pbc-0.5.14.tar.gz
tar -xvf pbc-0.5.14.tar.gz
cd pbc-0.5.14
./configure ; make ; sudo make install
export LIBRARY_PATH=$LIBRARY_PATH:/usr/local/lib
export LD_LIBRARY_PATH=$LD_LIBRARY_PATH:/usr/local/lib
charm
sudo apt-get install python3-dev
git clone https://github.com/JHUISI/charm.git
cd charm
git checkout 2.7-dev
sudo python setup.py install
pycrypt
sudo python -m pip install pycrypto
Clone the code:
git clone https://github.com/amiller/HoneyBadgerBFT.git
git checkout another-dev
Generate the keys
-
Threshold Signature Keys
python -m HoneyBadgerBFT.commoncoin.generate_keys N (t+1) > thsigN_t.keys
-
ECDSA Keys
python -m HoneyBadgerBFT.ecdsa.generate_keys_ecdsa N > ecdsa.keys
Threshold Encryption Keys
python -m HoneyBadgerBFT.threshenc.generate_keys N (N-2t) > thencN_t.keys
Usually, we run ecdsa key generation with large N just once because it can be re-used for different N/t. And we can store threshold signature keys and threshold encryption keys into different files for convenience.
python -m HoneyBadgerBFT.test.honest_party_test -k thsigN_t.keys -e ecdsa.keys -b B -n N -t t -c thencN_t.keys
Notice that each party will expect at least NlgN many transactions. And usually there is a key exception after they finish the consensus. Please just ignore it.
At HoneyBadger/ec2/ folder, run
python utility.py [ec2_access_key] [ec2_secret_key]
In this interactive ipython environment, run the following:
-
Prepare the all the keys files and put them in your local directory (namely ec2 folder)
(See the instructions above)
-
Launch new machines
launch_new_instances(region, number_of_machine) -
Query IPs
ipAll() -
Synchronize keys
c(getIP(), 'syncKeys') -
Install Dependencies
c(getIP(), 'install_dependencies') -
Clone and repo
c(getIP(), 'git_pull') -
Launch the experiment
c(getIP(), 'runProtocol:N,t,B')
where N, t, B are experiment parameters (replace them with numbers).
-
Implement distributed key generation
-
Investigate better parameterization and add support for larger key sizes
-
Replace plain TCP sockets with reliable/authenticated channels
-
Integration with Hyperledger, Open Blockchain, etc.
Interested in contributing to HoneyBadgerBFT? Developers wanted. Contact ic3directors@systems.cs.cornell.edu for more info.