Fyi, I love your content in general, but this twitter design article misses the use case/functional breakdown part, which is the thing I love the most about your content (like your book).
Context: I am a product manager and I read your book to understand how to do function/api/system design based on high-level product solution (e.g. eCommerce payment -> pay in and pay out). There are a lot of internet articles like this twitter article (which feels like architecting) but there are very few article about functional design (or at least I don't know how to find them, searching by keywords is quite futile)
There were 3 main challenges while scaling Twitter timeline back then:
> 150M world wide active users
-> 300k QPS for timelines
-> naive timeline "materialization" can be slow
The tech talk is very good but it is 40-min long. I have to cut out many details to fit in a 5-min read. To gain more context, please check the video mentioned earlier.
Big fan of your work and book. I've been reading your blogs regularly.
One request: Can you please also add the sources for the content. It would be helpful to explore that in detail.
Thank you. That's a great suggestion.
Fyi, I love your content in general, but this twitter design article misses the use case/functional breakdown part, which is the thing I love the most about your content (like your book).
Context: I am a product manager and I read your book to understand how to do function/api/system design based on high-level product solution (e.g. eCommerce payment -> pay in and pay out). There are a lot of internet articles like this twitter article (which feels like architecting) but there are very few article about functional design (or at least I don't know how to find them, searching by keywords is quite futile)
Thank you for the feedback. The architecture is from a Twitter tech talk given in 2013. https://www.infoq.com/presentations/Twitter-Timeline-Scalability/
There were 3 main challenges while scaling Twitter timeline back then:
> 150M world wide active users
-> 300k QPS for timelines
-> naive timeline "materialization" can be slow
The tech talk is very good but it is 40-min long. I have to cut out many details to fit in a 5-min read. To gain more context, please check the video mentioned earlier.