BlockRaFT is a crash-tolerant and scalable distributed framework designed to improve the availability and performance of blockchain nodes, particularly in permissioned settings. Instead of replicating full nodes per organization, BlockRaFT distributes the internal workload of a single blockchain node across a cluster using a RAFT-based leader–follower architecture. This design preserves fairness in consensus while enabling scalability and resilience to node crashes. Additionally, the framework introduces a concurrent Merkle tree optimization that decouples smart contract execution from state updates, significantly reducing execution overhead. Experimental results demonstrate substantial performance improvements over traditional single-core and multi-core baselines while maintaining strong fault tolerance.

To evaluate BlockRaFT, we compare our distributed framework against two single-node baselines: a serial scheduler and a parallel multi-core scheduler. The goal is to isolate the benefits of distributed workload partitioning and crash tolerance beyond local parallelism.

BlockRaFT Distributed Framework

A C++-based distributed framework built with CMake, supporting REST APIs, clients, DAG-based execution, protobuf-based communication, and coordination via etcd and Redpanda.

---

The repository is organized as follows:

ICSA_ARTIFACT/
├── BlockRaFT-distributed_node --> BlockRaFT
├── singleNode-parallel        --> Parallel execution mode
├── singleNode-serial          --> Serial execution mode
└── Experiment_files           --> All necessary exp related files

---

1. Setup

The initial setup requires more than 30 minutes, primarily due to the following steps:

1. Provisioning and configuring multiple virtual machines (VMs) to emulate a distributed cluster environment.
2. Installation and configuration of etcd for distributed coordination and RAFT-based leader election.
3. Installation of required dependencies including: Google Protocol Buffers (protobuf) for serialization.
4. Google Test (gtest) for unit testing.

The setup time may vary depending on system configuration, network bandwidth, and VM provisioning speed. The framework supports two deployment configurations:

1. Single Node Framework
2. BlockRaFT distributed Node Framework

---

1.1 Single Node Framework

This configuration runs the framework on a single machine using one VM (nodeS).

---

1.1.1 Requirements

Ensure the following are installed:

• CMake >= 3.20
• GCC >= 11.4.0
• Google Protocol Buffers
• Boost
• gRPC
• RocksDB
• etcd client libraries (if required)
• Multipass

---

1.1.2 Install & VM Setup

Install Multipass (Ubuntu)

The VM is created using Multipass, and all dependencies are installed inside the VM. After building via CMake, correctness can be verified using ctest. aq

Install snap in case not availbale

sudo apt update
sudo apt install snapd

sudo snap install multipass

Verify installation:

sudo multipass version

Create Single Node VM

sudo multipass launch -n nodeS --cpus 8 --mem 10G --disk 20G 24.04

Enter the VM:

sudo multipass shell nodeS

Install Dependencies Inside VM

sudo apt-get update && \
sudo apt-get install -y cmake unzip python3-pip protobuf-compiler libprotobuf-dev \
  libboost-all-dev libssl-dev libcpprest-dev libboost-system-dev \
  libcurl4-openssl-dev librdkafka-dev libasio-dev libgflags-dev \
  nlohmann-json3-dev libgtest-dev librocksdb-dev \
  libgrpc-dev libgrpc++-dev protobuf-compiler-grpc etcd-client libtbb-dev

---

1.1.3 Execution Modes

After completing the above steps, your Single Node environment is ready.

Copy the required module folder depending on the execution mode:

• Serial Execution → transfer to nodeS:singleNode-serial
• Parallel Execution → transfer to nodeS:singleNode-parallel

Example:

cd singleNode-serial
sudo multipass transfer -r ./ nodeS:singleNode-serial

OR,

cd singleNode-parallel
sudo multipass transfer -r ./ nodeS:singleNode-parallel

---

1.1.4 Building the Project

Inside the VM:

For Serial execution mode :

cd singleNode-serial
mkdir -p build
cd build
cmake ..
make

For Parallel execution mode :

cd singleNode-parallel
mkdir -p build
cd build
cmake ..
make

For verify the proper setup :

cd build 
ctest

---

1.2 BlockRaFT Distributed Node Framework

This configuration runs the framework across multiple VMs (node0, node1, node2) using etcd and Redpanda.

---

1.2.1 Project Structure

Distributed_Framework/
├── build/
├── blocksDB
├── block_producer
├── DAG-module
├── merkelTree
├── Crow/
├── googletest/
├── protos/
├── client/
├── RestAPI/
├── auto_deploy.sh
└── CMakeLists.txt

---

1.2.2 Requirements

• CMake >= 3.20
• GCC >= 11.4.0
• Google Protocol Buffers
• Multipass

---

1.2.3 Installing Multipass (Ubuntu)

sudo snap install multipass

Verify installation:

sudo multipass version

---

1.2.4 Creating Multipass VMs

cd BlockRaFT-distributed_node
chmod +x autoDeploy.sh
sudo ./autoDeploy.sh

---

1.2.5 Copying Project Files to VMs

sudo multipass transfer -r ./ node0:Distributed_Framework
sudo multipass transfer -r ./ node1:Distributed_Framework
sudo multipass transfer -r ./ node2:Distributed_Framework

---

1.2.6 Installing System Dependencies

Enter each VM in three different terminals:

sudo multipass shell node0
sudo multipass shell node1
sudo multipass shell node2

Install dependencies inside each VM:

sudo apt-get update && \
sudo apt-get install -y cmake unzip python3-pip protobuf-compiler libprotobuf-dev \
  libboost-all-dev libssl-dev libcpprest-dev libboost-system-dev \
  libcurl4-openssl-dev librdkafka-dev libasio-dev libgflags-dev \
  nlohmann-json3-dev libgtest-dev librocksdb-dev \
  libgrpc-dev libgrpc++-dev protobuf-compiler-grpc etcd-client libtbb-dev

---

1.2.7 Installing rpk (Redpanda CLI)

curl -LO https://github.com/redpanda-data/redpanda/releases/latest/download/rpk-linux-amd64.zip && \
mkdir -p ~/.local/bin && \
export PATH="$HOME/.local/bin:$PATH" && \
unzip rpk-linux-amd64.zip -d ~/.local/bin/

---

1.2.8 Installing etcd-cpp-apiv3

git clone --recurse-submodules https://github.com/etcd-cpp-apiv3/etcd-cpp-apiv3.git
cd etcd-cpp-apiv3
mkdir build && cd build
cmake ..
sudo make -j4
sudo make install

---

1.2.9 Building the Project

Inside each VM:

cd Distributed_Framework
mkdir -p build
cd build
cmake ..
make

For verify the proper setup :

cd build 
ctest

---

---

2. Experiment Configuration & Running

This section explains how to configure and reproduce experiments.

---

2.1 Configuration File

Each execution folder contains:

config.json

Example:

{
  "threadCount": 8,
  "txnCount": 10000,
  "blocks": 2,
  "scheduler": "parallel",
  "mode": "production"
}

Parameters

• threadCount → Worker threads (parallel mode)
• txnCount → Number of transactions
• blocks → Blocks generated
• scheduler → "serial" or "parallel"
• mode → Execution mode

---

2.2 Experiment Workloads

Located in:

ICSA_ARTIFACT
└──Experiment_files/
   ├── Wallet/
   └── Voting/

File Naming Format

<TxnCount>-<ConflictPercentage>.txt

Examples:

• 1000-0.txt
• 5000-50.txt

---

2.3 Preparing an Experiment

Step 1: Select Workload

Choose from:

Experiment_files/Wallet/

or

Experiment_files/Voting/

---

A. Single Node Experiments

Step 2: Copy Transactions

Copy into:

singleNode-serial/testFile/testFile.txt

or

singleNode-parallel/testFile/testFile.txt

inside selected execution folder.

B. BlockRaFT - Distributed Node Experiments

Step 2: Copy Transactions

Copy into:

BlockRAFT-distributed_node/leader/testFile.txt

inside selected execution folder.

---

Step 3: Copy Setup File

Copy:

Experiment_files/Wallet/setup.txt

or

Experiment_files/Voting/setup.txt

Copy into:

singleNode-serial/testFile/setup.txt

or

singleNode-parallel/testFile/setup.txt

inside selected execution folder.

B. BlockRaFT - Distributed Node Experiments

Step 3: Copy Setup File

Copy:

Experiment_files/Wallet/setup.txt

Copy into:

BlockRAFT-distributed_node/leader/setupFile.txt

inside selected execution folder.

---

2.4 Running Experiments

---

Single Node — Serial

cd singleNode-serial/build
./node

---

Single Node — Parallel

cd singleNode-parallel/build
./node

BlockRaFT-distributed node

Run the below mentioned command in each VMs for example for 3 Node setup i.e (node0, node1, node2).

cd BlockRAFT-distributed_node/build
./node

---

2.2.4 Cleaning Up

sudo multipass stop node0 node1 node2
sudo multipass delete node0 node1 node2
sudo multipass purge

---

3. Reproducibility Guidelines

To reproduce results reported in the paper:

• Use identical config.json parameters
• Use the same workload file
• Use identical VM specs
• Avoid background workloads
• Run multiple iterations and report averages

---

Notes

• All distributed nodes must use identical builds.
• Ensure etcd and Redpanda ports are open.
• Maintain consistent startup order.
• Recommended VM: 8 CPUs, 10GB RAM.