lillo / lmcs-analysis-validation

The repository contains the scripts and data for the experimental evaluation described in the paper "A theory of transaction parallelism in blockchains"

README - A theory of transaction parallelism in blockchains

This repository contains the data and the scripts used for the experimental evaluation described in the paper

M. Bartoletti, L. Galletta and M. Murgia. A theory of transaction parallelism in blockchains, Log. Methods Comput. Sci. 17(4), 2021.

Tools used for the experiment

• Go Ethereum v1.10.3-stable-991384a7
• Truffle v5.3.3
• Python 3
• Commandline dependency builder

Structure of the repository

Here is a high level description of the content of the repository.

.
├── contracts/
│   Source code of the Lottery contract. 
│
├── migrations/
│   Truffle scripts for the migration and deployment of the contract on the blockchain.
│
├── scripts/
│   Scripts used for the generation and analysis of the experiment data   
│
├── experiment-data
│   Data used for the evaluation
│
├── LICENSE
│
├── truffle-config.js
│   Truffle configuration.
│
└── README.md

Scripts

Here is a high level description of scripts used for the evaluation.

scripts
├── create-account.js
│   Script for the geth console to create further two accounts to play the lottery    
│
├── run-lottery.sh
│   Script bash that creates the required accounts, deploys the contract on the blockchain and plays 70 game of the lottery. It stores the hashes of the sent transactions into a file.
│
├── transactions-lottery.js
│   Truffle script that plays the lottery interacting with the contract on the blockchain.   
│
├── analyze-result.py
│   Script python that extracts the time spent by each transaction from the geth log, that computes the sequential time and that generates the input for cldb tool 
│
└── longest-path.py 
    Script python that given the dependency graph generated by cldb and the execution time of each transaction   computes a parallel schedule as the longest and most expensive path (in terms of time) of the dependency graph. 

Data

Here is a high level description of data generated for the evaluation.

experiment-data
├── geth-run[1-10].log
│   Log generated by geth 
│
├── lottery-run[1-10].log
│   Hashes of transaction generated during the game
│
├── lottery-info.json
│   Template used to generate the block to use as input of cldb
│
├── lottery-generated-block.json
│   The block of transactions used in our evaluation  
│
├── lottery-transaction-cost.txt
│   The cost in term of time of each transaction in lottery-generated-block.json 
│
├── lottery-transaction-graph.txt
│   The dependency graph generated by cldb for transactions in lottery-generated-block.json    
│
└── cldb-time.txt 
    The average time spent by cldb to perform the analysis of transaction in lottery-generated-block.json 

Steps of the experiment

Note steps 1 and 2 below are repeated 10 times.

1. Run geth creating a local development chain and storing the logs into a file

$ geth --dev --http --allow-insecure-unlock 2> geth_run.log

2. Deploy the Lottery smart contract and play the game

$ bash scripts/run-lottery.sh lottery-run.log

where lottery-run.log will store info about the called functions and their corresponding transactions. The script will ask password for the new accounts, the empty password is just fine for our purposes.

3. Extract the transaction execution time and generate the transaction block

$ python scripts/analyze-result.py lottery-run geth_run 10 lottery-info.json

where 10 is the number of repetitions of steps 1 and 2. The above script generates the file transaction-cost.txt and generated-block.json.

4. Generate the dependency graph

$ cldb generated-block.json

The script above will generate the file transaction-graph.txt.

5. Compute and longest and most expensive path in the dependency graph

$ python scripts/longest-path.py transaction-graph.txt transaction.cost.txt