Hacker News new | past | comments | ask | show | jobs | submit login

Why did it matter that a frame was transmitted in 2^12 microseconds, though? Why not 2^11 or 2^13?



2^12 wasn’t a constraint. The constraint was the clock speed of 3MHz. But making the frame size 1500 Bytes was nice because you get a clean power of two for frame transmission time.

You could lower or raise the frame size and have it take any amount of microseconds you want. Perhaps ~4ms was selected with human interaction times in mind. You can send a frame, process for a few ms, then send one frame back quick enough for feedback to appear as instant.


And if a second was defined centuries ago to be longer or shorter, we'd end up with some other length of time that was a power of 2 close to 4ms, resulting in a different MTU.

If that's indeed the reasoning, then it's amazing how arbitrary decisions made in the past end up deciding today's standards.

Now I want the story of the 9000 bytes MTU.


> Now I want the story of the 9000 bytes MTU.

* 2018: https://blog.dave.tf/post/jumbo-frames/

* April 2023: https://news.ycombinator.com/item?id=35647779




Consider applying for YC's Spring batch! Applications are open till Feb 11.

Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: