TCP Sucks • Hacker News Board

Re: TCP Sucks

#61

Post by berbec » Wed, May 06, 2020, 1:27 AM UTC

nine_k wrote: post #54

TheDong wrote: post #16

Earlier quoted context omitted.

Coal power obviously doesn't suck as its pretty much powered the planet for the last 100 years. It's perfectly possible for something to both be incredibly popular and to suck. An appeal to popularity is certainly not better than a click-bait headline. I'd argue that most of the stuff we use sucks in some ways as well; suck isn't an absolute term. TCP sucks in very specific ways the article talks about.

It's not just popularity; it's an adaptable enough design to work in the range from 150 baud acoustic modem links to 40 Gbit fiber optics. Not many designs can withstand that, no matter how hard people try.

I'd suggest an update to your lower bound - avian carrier

Re: TCP Sucks

#62

Post by the8472 » Wed, May 06, 2020, 11:43 AM UTC

berbec wrote: post #61

nine_k wrote: post #54

Earlier quoted context omitted.

It's not just popularity; it's an adaptable enough design to work in the range from 150 baud acoustic modem links to 40 Gbit fiber optics. Not many designs can withstand that, no matter how hard people try.

I'd suggest an update to your lower bound - avian carrier

To my knowledge that has only ever been deployed to send ICMP messages, not TCP. With a TCP stack tuned to give more lenient timeouts it might work but you'll need a lot more trips to do a whole connection setup and teardown.

Re: TCP Sucks

#63

Post by alyrik » Wed, May 06, 2020, 2:44 PM UTC

Animats wrote: post #49

klodolph wrote: post #13

Earlier quoted context omitted.

Complaining about the article’s title is no substitute for replying to the article’s content. Writing a provocative title is how you get people to read your article. If you have an opinion to share, you shouldn’t try to disguise it as a drab technical report. The article makes sound points and the title makes sense. Compare “TCP Sucks” (easy to read—communicates, concisely, that this is an opinion piece, and what the…

The author says "TCP sucks", and is promoting his own protocol. Iridium. Which he hasn't even designed yet . That's a bit much. We don't get to criticize his approach. (Also, if you're designing a communications system, don't give it the same name as another communications system.) TCP has its limitations, of course. It was designed to work over a wide range of connections, including dial-up, and it does. It's subopt…

> The author is worried about the overhead of the three-way handshake, but the overhead of setting up TLS is far worse.

Oh, "TLS sucks" is a whole separate blog post for another day ;-)

Re: TCP Sucks

#64

Post by fanf2 » Wed, May 06, 2020, 5:30 PM UTC

yaur wrote: post #55

TCP is fine. There is some questionable things here and I'll just touch on a couple of them. 1) TCP checksums are a bit weak. The author claims that we don't know how many errors are not caught by TCP's CRC and while that's true on some level the probability of bit errors is driven by properties of the physical medium over which the packet is traveling, not the transport layer protocol in use. That's why most, if not…

It's weird that the article doesn't mention SACK which is pretty much supported everywhere these days and reduces the head-of-line blocking problem significantly.