A tweet is 300 bytes. Why does storage hit petabytes?

transcript

A tweet is about 300 bytes. Tiny. So why does a Twitter-like app need petabytes a year?

100 million tweets a day, at 300 bytes. 30 gigabytes. Sounds like a lot? One tweet in ten carries a 200 kilobyte image. 2 terabytes a day. One in a hundred has a 5 megabyte video. 5 terabytes a day. Look at the text bar. It's all but vanished.

Together, about 7 terabytes of new storage a day. Times 365. 2 point 5 petabytes a year. And media is more than 99 percent of it. The tweets are a rounding error.

But here's the catch. Those images also go out, to every reader. Peak egress, 30 gigabits a second. All from your origin. 15 on an average day. Put a CDN in front, serving 90 percent of image requests. The origin drops to 2 point 2.

That's the step people skip. The so what. 30 gigabits isn't the answer. Images go through a CDN is.

Estimate the text, and you'll miss the system. Size the media. Then let the number make the decision.

Full breakdown in the lecture.