ReliaDL: Production-Grade Resilient Download Engine

ReliaDL is a production-grade, fault-tolerant parallel file download framework and high-throughput streaming engine. Designed for high-reliability data pipelines, enterprise infrastructure, and non-stationary channels, ReliaDL combines stochastic network optimization, per-chunk cryptographic integrity verification, corporate proxy tunneling, async rate limiting, and Prometheus observability.

PyPI Package | Documentation | Release Notes | RFC Roadmap | Issue Tracker

Key Capabilities

Installation

From PyPI (Recommended)

pip install reliadl

From Source

git clone https://github.com/PxA-Labs/ReliaDL.git
cd ReliaDL
pip install -e .

Verification

To verify that ReliaDL is correctly installed and ready to use, run these quick test commands in your terminal:

# 1. Verify package import and installation path
python -c "import reliadl; print('ReliaDL successfully imported from:', reliadl.__file__)"

# 2. Inspect package version and metadata
pip show reliadl

# 3. Test core component initialization
python -c "from reliadl import DownloadConfig; config = DownloadConfig(); print('Config initialized successfully!')"

Quick Start & Python SDK

1. Basic File Download & Resume Engine

import asyncio
from src.state_manager import StateManager

# Initialize resilient transfer state
state_mgr = StateManager(target_path="./downloads/large_dataset.tar.gz")
print(f"Transfer state initialized: {state_mgr}")

2. Corporate Proxy Tunneling (SOCKS5 & HTTP CONNECT)

from src.adapters.proxy_adapter import ProxyConfig, ProxyType, ProxyTunnel

# Configure SOCKS5 proxy with remote DNS resolution
proxy_config = ProxyConfig(
    proxy_type=ProxyType.SOCKS5H,
    host="proxy.corp.internal",
    port=1080,
    username="service_account",
    password="secure_password"
)

# Open secure tunnel to destination endpoint
tunnel = ProxyTunnel(proxy_config, timeout=30.0)
connection = tunnel.open("secure.example.com", 443)

3. Bandwidth Rate Limiting (Token Bucket)

import asyncio
from src.rate_limiter import TokenBucketRateLimiter

async def main():
    # Throttle transfer rate to 10 MB/s with a 2 MB burst capacity
    limiter = TokenBucketRateLimiter(rate=10 * 1024 * 1024, capacity=2 * 1024 * 1024)
    
    # Compute and acquire bandwidth delay before downloading next chunk
    delay = await limiter.acquire(64 * 1024)
    print(f"Paced chunk delay: {delay:.4f} seconds")

asyncio.run(main())

4. Prometheus Telemetry Monitoring

from src.telemetry.metrics import DownloadMetrics, MetricsRegistry, MetricsServer

registry = MetricsRegistry()
metrics = DownloadMetrics(registry)

# Record chunk transfer telemetry
metrics.record_chunk(bytes_count=65536, duration_seconds=0.012, mirror="us-east-1")

# Launch background HTTP metrics server for Prometheus scrapers on port 9090
with MetricsServer(registry, port=9090) as server:
    print("Prometheus scraper active at http://localhost:9090/metrics")

5. Cryptographic Stream Hash Verification

from src.hash_verifier import StreamingHashVerifier, constant_time_compare

# Initialize streaming SHA-256 verifier
verifier = StreamingHashVerifier(algorithm="sha256")
verifier.update(b"chunk byte payload data...")

# Compute digest and compare in constant time
digest = verifier.hexdigest()
is_valid = constant_time_compare(digest, "b94d27b9934d3e08a52e52d7da7dabfac484efe37a5380ee9088f7ace2efcde9")
print(f"Payload valid: {is_valid}")

Command Line Interface (CLI)

ReliaDL includes a production CLI (reliadl or python -m src.main) for automated transfers, system diagnostics, pre-flight probing, telemetry monitoring, and Merkle tree auditing:

# 1. Parallel File Download
reliadl download --url "https://example.com/dataset.tar.gz" --output "./downloads/dataset.tar.gz" --adachunk --whittle

# 2. High-Throughput Network & Crypto Benchmark
reliadl benchmark --duration 10s --block-size 64

# 3. Pre-Flight Infrastructure Diagnostic Probe
reliadl probe --url "https://cdn.example.com/data.tar.gz" --proxy "socks5h://proxy.corp:1080"

# 4. State Checkpoint & Manifest Diagnostic Inspection
reliadl inspect-state --state-file "./downloads/.reliadl/dataset.tar.gz.state" --json

# 5. System Environment & Storage Diagnostic Doctor
reliadl doctor

# 6. Real-Time Telemetry UI Monitoring Dashboard
reliadl top --metrics-port 9090

# 7. Segment Merkle Tree Cryptographic Audit
reliadl hash-tree --file "./downloads/dataset.tar.gz" --merkle-root "a3f8..."

# 8. Verification & Resumption
reliadl verify --file "./downloads/dataset.tar.gz" --expected-hash "sha256:e3b0c442..."
reliadl resume --state-file "./downloads/.reliadl/dataset.tar.gz.state"

System Architecture

graph TB
    subgraph ClientLayer["1. Client & Configuration Layer"]
        CLI["CLI Interface (src.main)"]
        CONF["Configuration Engine (src.config)"]
        SM["ACID State Persistence (src.state_manager)"]
    end

    subgraph ControlLayer["2. Optimization & Network Layer"]
        AC["AdaChunk Optimizer (src.algorithms.adaptive_chunker)"]
        RL["Token Bucket Rate Limiter (src.rate_limiter)"]
        PA["Proxy Tunnel Adapter (src.adapters.proxy_adapter)"]
    end

    subgraph TransportLayer["3. Transport Engine & Worker Pool"]
        DE["Download Engine (src.download_engine)"]
        W1["Worker 1 (HTTP/2 Range GET)"]
        W2["Worker 2 (HTTP/2 Range GET)"]
        WN["Worker K (HTTP/2 Range GET)"]
    end

    subgraph SecurityLayer["4. Cryptographic Verification & Observability"]
        DH["Dual Hasher (src.hash_verifier)"]
        ML["Sub-Chunk Merkle Localizer"]
        TEL["Prometheus Metrics Server (src.telemetry)"]
    end

    subgraph StorageLayer["5. Zero-Copy Positional IO"]
        FA["Positional Disk Writer (src.file_assembler)"]
        OUT["Target Payload Artifact"]
    end

    CLI --> CONF
    CLI --> DE
    DE <--> SM
    DE --> AC
    DE --> RL
    DE --> PA
    PA --> W1 & W2 & WN
    W1 & W2 & WN --> DH
    DH --> TEL
    DH --> ML
    DH --> FA
    FA --> OUT

Repository & Module Structure

ReliaDL/
├── README.md                          # Project documentation entry point
├── pyproject.toml                     # PEP 517/518 PyPI build configuration
├── LICENSE                            # Apache 2.0 License
├── requirements.txt                   # Dependency manifests
│
├── .github/
│   ├── dependabot.yml                 # Automated weekly security updates
│   └── workflows/
│       ├── codeql.yml                 # CodeQL static security analysis
│       ├── ci.yml                     # Continuous integration test matrix
│       └── publish.yml                # PyPI Trusted Publisher (OIDC) workflow
│
├── docs/                              # Architecture documentation & research specifications
│   ├── NOVEL_ALGORITHMS.md            # Formal mathematical formulations & proofs
│   ├── ARCHITECTURE.md                # System design trade-offs
│   ├── API_REFERENCE.md               # API documentation
│   └── OBSERVABILITY.md               # Prometheus telemetry guide
│
├── src/                               # Core engine modules
│   ├── adapters/                      # Transport & proxy adapters
│   │   ├── proxy_adapter.py           # SOCKS5 & HTTP CONNECT corporate tunneling
│   │   ├── s3_adapter.py              # AWS S3 cloud adapter
│   │   ├── gcs_adapter.py             # Google Cloud Storage adapter
│   │   └── azure_adapter.py           # Azure Blob Storage adapter
│   ├── algorithms/                    # Stochastic optimization algorithms
│   │   ├── adaptive_chunker.py        # AdaChunk Lyapunov optimizer
│   │   ├── metrics_collector.py       # EWMA network estimator
│   │   ├── mirror_bandit.py           # Whittle index bandit scheduler
│   │   └── work_stealer.py            # Dynamic work stealing queue
│   ├── hash_verifier.py               # Streaming SHA-256 & LtHash verifier
│   ├── rate_limiter.py                # Token bucket bandwidth rate limiter
│   ├── telemetry/                     # Prometheus telemetry & exporter
│   ├── state_manager.py               # Atomic state file manager
│   ├── file_assembler.py              # Zero-copy positional disk assembler
│   ├── config.py                      # System configuration parser
│   ├── exceptions.py                  # Custom taxonomy exceptions
│   └── main.py                        # Command Line Interface (CLI)
│
└── tests/                             # Automated test suite (860+ tests passing)
    └── unit/                          # Component unit tests

Security, Governance & Compliance

ReliaDL adheres to strict enterprise security standards and the Open Source Security Foundation (OpenSSF) guidelines:

• CodeQL Static Security Analysis: Continuous automated scanning for security vulnerabilities on every push and pull request.
• Dependabot Vulnerability Management: Weekly automated dependency updates for pip packages and GitHub Actions.
• PyPI Trusted Publisher (OIDC): Tokenless, passwordless release publishing using OpenID Connect and GitHub Artifact Attestations.
• Cryptographic Supply Chain Verification: SHA-256 digest validation and signed manifest catalogs preventing payload tampering.
• Vulnerability Reporting: Coordinated security disclosure policy detailed in SECURITY.md.

Distribution RFC & Multi-Phase Roadmap

ReliaDL is executing a multi-phase distribution roadmap discussed in RFC Discussion #83:

• Phase 1 (Core): PyPI Package (Issue #84) & MkDocs Documentation.
• Phase 2 (Binaries & Containers): Standalone Executables & GHCR Docker Image.
• Phase 3 (Package Managers): Homebrew / WinGet / Linux Package Managers.

Research Reference

If you use ReliaDL in academic research or production system studies, please cite:

@article{reliadl2026,
  title={{ReliaDL: Adaptive Fault-Tolerant Chunked Transfer with Homomorphic Verification and Stochastic Scheduling}},
  author={Joshi, Purvansh and Mittal, Archit},
  journal={arXiv preprint arXiv:2608.xxxxx},
  year={2026},
  publisher={PxA Labs},
  url={https://github.com/PxA-Labs/ReliaDL}
}

License

ReliaDL is open-source software licensed under the Apache License 2.0.