gcp-mft-ai

AI-powered Managed File Transfer for Google Cloud (GCS, Filestore, Storage Transfer Service)

gcp-mft-ai is an open-source, production-grade Python library that transforms traditional file transfers on Google Cloud Platform (GCP) into intelligent, ML-optimized, secure operations.

It automates, predicts, protects, and optimizes file movement across:

• Google Cloud Storage (GCS)
• Cloud Filestore (NFS-based File System)
• Storage Transfer Service (STS API)

Designed for large-scale enterprises, DevOps engineers, and AI/ML pipelines, gcp-mft-ai brings the future of AI-enhanced Managed File Transfer (MFT) into your cloud workflows.

Features

• Upload/download large files intelligently
• AES-256 encryption support
• Predict transfer time with ML
• Optimize best transfer windows
• Detect anomalies in transfer logs

Core Capabilities

• Multi-Service MFT: GCS bucket transfers, Filestore filesystem moves, GCP Storage Transfer Service orchestration
• Encryption at Source: AES-256-GCM authenticated encryption (optional per transfer)
• ML-Based Transfer Time Prediction: Predict upload/download times using Linear Regression or Random Forest models
• Anomaly Detection: Identify unusual slowdowns, spikes, or transfer errors automatically using Isolation Forest
• Transfer Window Optimization: Find the best network window (hour of day) to minimize congestion and maximize throughput
• Resilient Transfers: Automatic retries, resumable uploads for large objects, GCP API throttling handling
• Config-Driven Automation: Manage all settings via simple YAML or JSON configuration files

Internal Architecture

• GCS Transfers: Built atop the google-cloud-storage SDK for resumable, secure, and reliable object transfers.

• Filestore Transfers: Abstracted over NFS filesystem mounts, allowing simple shutil-based secure moves between instances or buckets.

• Storage Transfer Service API: Dynamically creates and monitors cloud-to-cloud transfer jobs via authenticated REST API calls (fully IAM compliant).

• Prediction Engine: i) Trained on historical transfer data (file_size_mb, transfer_time_sec), supports: ii) Linear Regression (lightweight, fast) iii) Random Forest (higher-accuracy, non-linear patterns)

• Anomaly Detection: Isolation Forest model isolates unusual file size vs time behavior — flagging spikes, failures, and risks early.

• Encryption Layer: AES-256 encryption with GCM mode ensures data integrity and confidentiality before movement.

• Optimization Layer: Hour-by-hour analysis of historical transfer speeds to recommend the best operational windows.

Security-First Design

• Encryption: Native AES-256-GCM encryption/decryption for any file before or after cloud storage.

• Token Management: Secure OAuth2 token usage for Storage Transfer Service API access.

• No plaintext secrets: Designed for service account usage via environment or config.

Usage Overview

• Upload to GCS: upload_to_gcs(source_path, bucket, destination_path)
• Download from GCS: download_from_gcs(blob_name, bucket, destination_path)
• Upload to Filestore: upload_to_filestore(source_path, mount_point, relative_path)
• Launch Storage Transfer Job: launch_storage_transfer_job(source_bucket, destination_bucket, project_id)
• Predict Transfer Time: predict_transfer_time(file_size_mb)
• Detect Anomalies: detect_transfer_anomalies(csv_log_path)
• Find Best Transfer Window: find_best_transfer_window(csv_log_path)

Real-World Use Cases

• Media & Entertainment: Migrate large UHD videos to GCS for editing pipelines
• AI/ML Model Training: Transfer terabyte datasets securely and predictably to TPU training zones
• Backup & Disaster Recovery: Automate and encrypt cross-region backup uploads with anomaly alerting
• Healthcare & Finance: Securely move critical records across cloud environments with end-to-end encryption
• Retail Analytics: Optimize massive log file ingestion pipelines to GCP data lakes

Technology Stack

• Python 3.7+

• Google Cloud SDKs (google-cloud-storage, requests)

• Cryptography (AES-256-GCM secure cipher)

• scikit-learn (ML Models: Linear Regression, Random Forest, Isolation Forest)

• pandas (Data preparation for ML)

• pyyaml (Config loading)

• joblib (Model persistence)

MIT License

Author: Raghava Chellu

MIT License is freely usable for academic, personal, and commercial projects.

Installation

pip install gcp-mft-ai

Deployment Readiness

• PyPI-ready (setup.py, pyproject.toml)

• Full unit testing (unittest framework)

• Full documentation (README.md, examples/)

• Cloud deployment friendly (Docker/CI/CD pipelines)

Conclusion

Traditional file transfers are simple. Modern file transfers must be intelligent, secure, and predictive. gcp-mft-ai brings cutting-edge AI and cloud-native automation to Managed File Transfer on Google Cloud securing your data, optimizing your operations, and helping you move smarter, stronger, and faster.