Home

Scalable Objects Persistence (SOP)

Scalable Objects Persistence (SOP) is an enterprise-grade, transactional storage engine designed for high-performance data management in Go and Python. It bridges the gap between raw file system speed and database reliability, offering a unified platform for Key-Value, Vector, and AI Model storage.

SOP eliminates the "impedance mismatch" between application objects and storage, allowing developers to persist complex data structures directly with full ACID guarantees.

🚀 Quick Access

Resource	Description
GitHub Repository	Source code, releases, and issue tracking.
Go Reference	Comprehensive API documentation for Go developers.
Python Bindings	Documentation and examples for sop4py.

---

📚 Documentation Center

Core Architecture

• Architecture Guide
  Explore the internal design, package structure, and backend storage options (infs vs incfs).
• Workflows & Scenarios
  Implementation patterns ranging from local standalone development to global-scale enterprise swarms.
• Swarm Computing
  Deep dive into SOP's distributed, masterless architecture for massive parallel processing.

Developer Guides

• Go API Cookbook
  Practical, copy-pasteable recipes for common implementation scenarios.
• Python Cookbook
  Specialized examples for Data Science and AI workflows, including Pandas integration and Vector Search.
• AI Expert System Tutorial
  Step-by-step guide to building a privacy-first "Doctor & Nurse" AI agent using SOP's Vector Store.
• Operational Guide
  Best practices for cluster deployment, monitoring, and disaster recovery.

---

🌟 Key Capabilities

Unified Storage Engine

• Multi-Model Support: Manage Vector Embeddings, AI Models, and Key-Value data within a single, cohesive system.
• ACID Compliance: Full transaction support (Begin, Commit, Rollback) with strict isolation, ensuring data integrity even during failures.

High-Performance Indexing

• Complex Keys: Define composite keys (e.g., Region -> Dept -> ID) using native structs/dataclasses. SOP handles multi-column sorting and indexing automatically.
• "Ride-on" Metadata: Embed critical metadata directly in the B-Tree key. This enables high-speed scanning and filtering of millions of records without the I/O penalty of fetching the full value payload.
• Vector Search: Built-in, transactional k-NN search for RAG (Retrieval-Augmented Generation) and similarity applications.

Enterprise Ready

• Hybrid Caching: Integrated Redis-backed L1/L2 caching for sub-millisecond access speeds.
• Multi-Tenancy: Native isolation support via Cassandra Keyspaces or Directory-based partitioning.
• Flexible Deployment: Seamlessly switch between Standalone (Local/Edge) and Clustered (Distributed) modes without changing application code.

---

🤝 Community & Support

• Discussions: Engage with the community, ask questions, and share use cases.
• Issues: Report defects or request new features.
• Contributing: Guidelines for contributing code and documentation.