Ask HN: What happened to Bufferbloat?

12 points·pjungwir··6 comments

A few years ago I heard a lot of people worrying about how large buffers in our routers would cause latency problems. It all sounded very serious---like the end of the Internet. But I haven't heard any news for a couple years. Everything on the Wikipedia page looks old:

https://en.wikipedia.org/wiki/Bufferbloat

Is this something people are still concerned about? Were the original reports just hyperbole?

View on HN

It is everywhere... pretty thoroughly quantified now (example:

http://www.dslreports.com/speedtest/results/bufferbloat?up=1

)

It is serious and remains so... but

Fixes have emerged (fq_codel, pie, cake) that are widely available in linux and linux derived gear -

which had spectacular results:

http://burntchrome.blogspot.fr/2014/05/fixing-bufferbloat-on...

and pie was mandated as part of the docsis 3.1 std.

and standardization efforts of the new aqms is taking place at the ietf. For example, this RFC just emerged: https://tools.ietf.org/html/rfc7567 and there are other products of that working group nearing finalization.

While wifi and lte remain to be fixed, I sleep better knowing that the tide is turning.

It didn't go away. I was reminded of it recently when I read this: https://www.febo.com/pipermail/time-nuts/2015-July/092770.ht...

The bufferbloat-related changes made in CeroWRT (http://www.bufferbloat.net/projects/cerowrt) have been merged into the Linux kernel and the OpenWrt project.

Seems like it's a phenomenon that goes through cycles of visibility, given its documentation dates back to the early 80s.

Twitch working on a form of buffer bloat recently: http://blog.twitch.tv/2015/05/new-reduced-stream-delay-beta/

(For clarity, not router buffer bloat, and there's no real technical discussion there either.)

Reply on news.ycombinator.com