Re: rfc791 coming up to 40 years ... what to do (remember, celebrate, ...?)

[Michael Thomas <[e-mail]>]

On 3/25/21 12:21 PM, Phillip Hallam-Baker wrote:
> On Wed, Mar 24, 2021 at 6:23 PM Keith Moore 
> <[e-mail] <mailto:[e-mail]>> wrote:
>
>      Not sure I agree that TCP was optimized for file transfer -
>     certainly a transport that's optimized for file transfer doesn't
>     need to ensure strict ordering of all messages sent from one peer
>     to another as long as they get reassembled in correct order when
>     the file is written.   Otherwise, I agree.
>
>     Keith
>
> The slow start parameters are.
>
> Sure, you can use TCP for many things, but if you are making an HTTP 
> GET request, slow start on send is doing the exact opposite of what 
> you would want (and often with parameters chosen for 1980s line speeds).
>
> But HTTP didn't exist when TCP was invented. There was terminal 
> connection mode, then proto-FTP emerged out of terminal mode and then 
> email from proto-FTP.
>
> My bigger point though is that don't try to make QUIC all things to 
> all people. It is better to hyper-optimise QUIC to Web browsing and 
> write a different transport for Web Services, a different protocol for 
> real time media, etc. etc.
>
>
Well to be fair, there was a lot of concern about congestive collapse of 
the internet back then. Considering we were all novices back in the 
80's, zillions of people rolling their own transport could have been a 
real mess. And of course targeting a small handful of kernels is a lot 
easier than convincing a zoo of different transports that they really 
ought to be doing this or that new bit of new research.

Mike