Course contentsShow
← System Design
Lesson 540 of 1,91915. Consistency ModelsFree lesson

Use Cases for Eventual Consistency

Identify systems that tolerate eventual consistency: DNS, CDNs, social feeds, shopping carts, and collaborative docs.

Use Cases for Eventual Consistency

What you'll learn: Which real-world systems can tolerate eventual consistency and why it's the right choice for them.


Eventual consistency isn't a compromise—it's often the optimal choice for systems where availability and performance matter more than instantaneous agreement across all nodes. Let's explore systems that benefit from this model.

Systems That Tolerate Eventual Consistency

DNS (Domain Name System)
DNS updates propagate gradually across servers worldwide. When you update a domain's IP address, different DNS servers may return different IPs for minutes or hours. This is acceptable because DNS changes are rare, and temporary inconsistency doesn't break the internet.

Content Delivery Networks (CDNs)
CDNs cache content at edge servers globally. When you update an image or webpage, different geographic regions might serve the old version until their cache expires. Users tolerate this because fast page loads matter more than seeing updates instantly.

Social Media Feeds
Your follower count or the exact ordering of likes doesn't need to be perfectly synchronized across all servers. If two users see slightly different like counts for a few seconds, the system still functions well. Availability and low latency trump perfect consistency here.

Shopping Carts
E-commerce sites prioritize letting you add items even if inventory counts are slightly stale. It's better to accept an order and potentially email about out-of-stock items later than to block purchases during network partitions.

Collaborative Documents
Google Docs allows multiple users to edit simultaneously. Your changes appear instantly to you, but others see them with slight delays. CRDTs and vector clocks ensure everyone eventually converges to the same document state without blocking writes.

The Common Thread

These systems share two traits: writes must always succeed (availability), and temporary inconsistency doesn't harm core functionality. Users understand and accept brief staleness in exchange for speed and reliability.

Key Takeaway: Eventual consistency excels in read-heavy, globally distributed systems where availability and low latency outweigh the need for instantaneous global agreement—DNS, CDNs, social feeds, shopping carts, and collaborative tools all make this tradeoff successfully.