---
title: "Comment by Rafat Islam on \"Introduction to system scalability and reliability\" | Hashnode"
description: "The problem with master slave architecture is consistency issues due to replication lag. What will happen when the master goes down without syncing some data to"
url: "https://hashnode.com/posts/introduction-to-system-scalability-and-reliability/60ce071e37952431a17126d6/comment/60f0e5ea19e3e845d50d05a7"
---

[Discussion](https://hashnode.com/posts/introduction-to-system-scalability-and-reliability/60ce071e37952431a17126d6)

Comment [Introduction to system scalability and reliability](https://hashnode.com/posts/introduction-to-system-scalability-and-reliability/60ce071e37952431a17126d6)

[Rafat Islam](https://hashnode.com/@Rafat)

[Jul 16, 2021](https://hashnode.com/posts/introduction-to-system-scalability-and-reliability/60ce071e37952431a17126d6/comment/60f0e5ea19e3e845d50d05a7)

The problem with master slave architecture is consistency issues due to replication lag. What will happen when the master goes down without syncing some data to the slaves? For that we might need to incorporate other strategy too like synchronous replication, but then again it'll add to the latency. Good read anyway. Please keep writing. Thank you. 😊

[Ahmed Sadman Muhib](https://hashnode.com/@ahmedsadman)

[Jul 16, 2021](https://hashnode.com/posts/introduction-to-system-scalability-and-reliability/60ce071e37952431a17126d6/comment/60f0e5ea19e3e845d50d05a7#reply-60f123a8303c5845ab8ca814)

No and No. Such operations are done asynchronously, whenever master receives write request it will pass them immediately to the replicas. Different solutions have different strategies to sync out of date data. For example, ElasticSearch uses Primary Term and Sequence Number in internal state to handle such issues.
