Skip to content

benchmark.md

Latest commit

 

History

History
102 lines (72 loc) · 4.4 KB

benchmark.md

File metadata and controls

102 lines (72 loc) · 4.4 KB

Benchmarking LND payments

The benchmark_receive.py and benchmark_send.py scripts measure the time it takes to receive or send a multi-part payment. This assumes that a direct channel is established between the node that is to be benchmarked and another node that idealy has its database stored on a local ramfs to minimize its impact on the benchmark results.

Prerequisites

Install Python 3.9 or newer.

Create a Python virtual environment

python3 -m venv lnd

Activate the virtual environment

source lnd/bin/activate

Install dependencies

pip install grpcio-tools

Copy the lightning.proto and router.proto files from the lnd repository or download them

curl -o lightning.proto -s https://raw.githubusercontent.com/lightningnetwork/lnd/master/lnrpc/lightning.proto
curl -o router.proto -s https://raw.githubusercontent.com/lightningnetwork/lnd/master/lnrpc/routerrpc/router.proto

Compile the proto files

python -m grpc_tools.protoc --proto_path=.  --python_out=. --grpc_python_out=. lightning.proto
python -m grpc_tools.protoc --proto_path=.  --python_out=. --grpc_python_out=. router.proto

Configure the benchmark

Create a file called benchmark.ini with the following contents:

[Common]
NUMBER_OF_SHARDS = 100
AMOUNT = 100
CHANNEL_ID = 113249697726465

[Target]
TARGET_IPPORT = 192.168.0.100:10009
TARGET_MACAROON = admin_alice.macaroon
TARGET_TLS_CERT = tls_alice.cert
TARGET_PUBKEY = 02fc8b1dea7f00153be3bd449cedaabe73fda8761c9fe568f4c3e035b766530161

[Peer]
PEER_IPPORT = 192.168.0.200:10009
PEER_MACAROON = admin_bob.macaroon
PEER_TLS_CERT = tls_bob.cert
PEER_PUBKEY = 02155577dea6afd9eb20471ac3e9a82fd4893e06b4ff7b22a3b706d381faca3eff

Adjust the configuration parameters according to the following explanation:

  • NUMBER_OF_SHARDS: The number of parts a payment should be split into. The more parts, the longer it takes to settle the payment.
  • AMOUNT: The number of sats per payment. This value must be divisible by NUMBER_OF_SHARDS. Changing the amount does not impact performance. It is recommended to set this to the same value as NUMBER_OF_SHARDS. (resulting in 1 satoshi per shard)
  • CHANNEL_ID: The id of the direct channel between the two nodes. Obtain it by running lncli listchannels | grep chan_id on either node.
  • TARGET_IPPORT: The IP address and port of the gRPC API of the node that is to be benchmarked. To expose the gRPC API externally, set rpclisten=0.0.0.0:10009 in the lnd.conf file of the node.
  • TARGET_MACAROON: The path to the admin.macaroon file, of the node that is to be benchmarked, found at .lnd/data/chain/bitcoin/<network>/admin.macaroon
  • TARGET_TLS_CERT: The path to the tls.cert file, of the node that is to be benchmarked, found at .lnd/tls.cert
  • TARGET_PUBKEY: The identity public key of the node that is to be benchmarked. It can be obtained by running lncli getinfo | grep identity_pubkey on the node that is to be benchmarked.
  • PEER_IPPORT: The IP address and port of the gRPC API of the peer node. To expose the gRPC API externally, set rpclisten=0.0.0.0:10009 in the lnd.conf file of the node.
  • PEER_MACAROON: The path to the admin.macaroon file of the peer node found at .lnd/data/chain/bitcoin/<network>/admin.macaroon
  • PEER_TLS_CERT: The path to the tls.cert file of the peer node found at .lnd/tls.cert
  • PEER_PUBKEY: The identity public key of the peer node. It can be obtained by running lncli getinfo | grep identity_pubkey on the peer node.

Run the benchmarks

To run one iteration of the receiving benchmark simply run the benchmark_receive.py script.

./benchmark_receive.py

When completed successfully, this appends the time in seconds it took for all HTLCs to settle to a file called benchmark_receive.log. Note that the time taken to write preimages to the peer node's database is not included, as it is not relevant to the benchmark.

To run multiple iterations of the benchmark, simply call the script in a loop:

while true; do ./benchmark_receive.py; done

This can be interrupted anytime with Ctrl + C.

To benchmark the sending performance, run the benchmark_send.py script.

while true; do ./benchmark_send.py; done

This measures the time from payment initiation until all preimages are successfully written to the database. The time in seconds is appended to a file called benchmark_send.log.