---
title: "Discussion on \"Crash-safe JSON at scale: atomic writes + recovery without a DB\" | Hashnode"
description: "Discussion on \"Crash-safe JSON at scale: atomic writes + recovery without a DB\". How we hardened file-based JSON storage in a Flask app - atomic writes (fsync+os.replace), rolling .bak, and forensic-safe recovery reads."
url: "https://hashnode.com/posts/crash-safe-json-at-scale-atomic-writes-recovery-without-a-db/695b7dc265de715b6e79b880"
---

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

Discussion

# [Crash-safe JSON at scale: atomic writes + recovery without a DB](https://c0nstanta.hashnode.dev/crash-safe-json-at-scale-atomic-writes-recovery-without-a-db?utm_source=hashnode&utm_medium=feed)

How we hardened file-based JSON storage in a Flask app - atomic writes (fsync+os.replace), rolling .bak, and forensic-safe recovery reads.

Jan 5 · 15 min read · c0nstanta.hashnode.dev  
Read the article

-   [#python](https://hashnode.com/tag/python)
-   [#json](https://hashnode.com/tag/json)
-   [#atomic-design](https://hashnode.com/tag/atomic-design)
-   [#reliability](https://hashnode.com/tag/reliability)

## Written by

[Konstantin](https://hashnode.com/@C0nstanta)

@C0nstanta

  
[View profile](https://hashnode.com/@C0nstanta)

Senior Software Engineer working on backend systems, data pipelines, and demand forecasting.

## Responses

Markdown supported · ⌘↵ to post

No responses yet.