Arpit's Newsletter read by 38000+ engineers
Weekly essays on real-world system design, distributed systems, or a deep dive into some super-clever algorithm.
GIPHY serves 10 billion GIFs every day, here’s how it beautifully uses different features of CDN.
Think of CDN as a geographically distributed cache; and just like any regular cache, it sits between the user and the origin.
For any request, if it has the data, it serves the response. If not, it hits the origin to grab the data, cache it, and then responds.
A key highlight of using a CDN is geographical nearness. Because the CDN servers are distributed worldwide, the request from a user is served from the nearest edge server giving an excellent UX.
This is a no-brainer application of CDN. Giphy serves all the media content like images and videos through CDN that sits transparently between the user and the origin (eg: S3).
Apart from the media content, Giphy uses CDN to cache API responses of Search and Discover APIs like
It serves these APIs from CDN because the responses of these APIs do not change often; hence using CDN for this reduces the load on API servers.
Not all APIs or media objects need to be cached on CDN for the same amount of time. Hence Giphy configures different expirations for different types of APIs.
Media object endpoints are cached longer while trending API is cached for a shorter duration.
Sometimes, it is the backend server that should dictate for how long the response should be cached.
Hence, Giphy, in the HTTP response from the origin server provides max-age
headers that tell CDN the TTL for the specific response. This gives finer control over key expiration.
Giphy uses Surrogate Keys (tags) while caching endpoints on CDN. It helps in smarter cache invalidation, eg:
Here's the video ⤵
Super practical courses, with a no-nonsense approach, are designed to spark engineering curiosity and help you ace your career.
An in-depth, self-paced, and on-demand course that for early engineers to become great at designing scalable, available, and extensible systems at scale.
A masterclass that helps experienced engineers become great at designing scalable, fault-tolerant, and highly available systems.
A course that helps covers Redis internals by reimplementing its core features like - event loop, serialization protocol, pipelining, eviction, and transactions.
Arpit's Newsletter read by 38000+ engineers
Weekly essays on real-world system design, distributed systems, or a deep dive into some super-clever algorithm.