Beyond Similarity Search: A Unified Data Layer for Production RAG Systems
Venkata Krishna Prasanth Budigi, Siri Chandana Sirigiri
TLDR
This paper proposes a unified PostgreSQL-based data layer for RAG systems, significantly improving reliability, performance, and security.
Key contributions
- Identifies root causes of RAG reliability issues: data staleness, tenant leakage, and query complexity explosion.
- Proposes a unified data layer for RAG systems using PostgreSQL with native pgvector and HNSW indexing.
- Achieves 92% latency reduction for date-filtered queries and 74% for tenant-scoped queries.
- Eliminates data leakage and sync inconsistencies, reducing synchronization code by 93%.
Why it matters
This paper addresses critical reliability issues in production RAG systems, offering a robust and efficient unified data layer solution. It significantly improves performance, security, and maintainability, making RAG more viable for real-world organizational knowledge grounding.
Original Abstract
Retrieval-Augmented Generation (RAG) systems have become the standard architecture for grounding large language models in organizational knowledge. Yet production deployments consistently expose a gap between clean prototype performance and real-world reliability. This paper identifies three root causes of that gap: data staleness, tenant data leakage, and query composition explosion. All three trace back to the conventional split-system data layer. We propose and evaluate a unified data layer built on PostgreSQL with native vector search (pgvector) and HNSW indexing. Controlled benchmarks on 50,000 documents show 92% latency reduction for date-filtered queries, 74% for tenant-scoped queries, zero synchronization inconsistency, and complete elimination of cross-tenant data leakage with 93% less synchronization code. We additionally discuss a recommended hybrid tier architecture
📬 Weekly AI Paper Digest
Get the top 10 AI/ML arXiv papers from the week — summarized, scored, and delivered to your inbox every Monday.