---
title: "Discussion on \"Scaling RAG Systems: Production Architecture, Performance, and Cost Optimization\" | Hashnode"
description: "Discussion on \"Scaling RAG Systems: Production Architecture, Performance, and Cost Optimization\". Part 4 of the Production RAG series: learn large-scale RAG architecture, vector DB performance, caching, async processing, and cost optimization."
url: "https://hashnode.com/posts/scaling-rag-systems-production-architecture-performance-and-cost-optimization/6a8577daead8bdbc4430b903"
---

[Feed](https://hashnode.com/feed) /

Discussion

# [Scaling RAG Systems: Production Architecture, Performance, and Cost Optimization](https://blog.damir-karimov.com/scaling-rag-systems-production-architecture-performance-and-cost-optimization?utm_source=hashnode&utm_medium=feed)

Part 4 of the Production RAG series: learn large-scale RAG architecture, vector DB performance, caching, async processing, and cost optimization.

Aug 19 · 22 min read · blog.damir-karimov.com  
Updated Aug 20  
Read the article

-   [#rag](https://hashnode.com/tag/rag)
-   [#llm](https://hashnode.com/tag/llm)
-   [#ai](https://hashnode.com/tag/ai)
-   [#vector-database](https://hashnode.com/tag/vector-database)
-   [#software-architecture](https://hashnode.com/tag/software-architecture)

## Written by

[Damir Karimov](https://hashnode.com/@damir-karimov)

@damir-karimov · Tashkent, Uzbekistan

  
[View profile](https://hashnode.com/@damir-karimov)

Senior Fullstack & Frontend Developer | React, Node.js, Python | 8+ yoe | AI & Automation

Software engineer focused on React, TypeScript, and Next.js ecosystems. Designs scalable frontend architectures (FSD), real-time systems, and backend integrations. Builds automation workflows and AI-driven features for production-grade web platforms.

## Responses 1

Markdown supported · ⌘↵ to post

[Kartik N V J K](https://hashnode.com/@kartik-nvj)

[Aug 21](https://hashnode.com/posts/scaling-rag-systems-production-architecture-performance-and-cost-optimization/6a8577daead8bdbc4430b903/comment/6a88a1e09aa265616830112a)

The cost section resonates: retrieval quality and cost pull in opposite directions once you scale the index. I found reranking let me pull fewer candidates up front without losing recall, which dropped both latency and spend. Are you caching embeddings at the query level, or only on the document side?