Great #QconNYC 21st century languages track thanks to @kate_sills @FiloSottile @steveklabnik @ag_dubs @funcOfJoe for your amazing talks!
-
Show this thread
-
I felt the 21st century theme really worked. Doing things thoughtfully with standards and consensus while trying to get them right was a theme. Languages and ways of using them and open design have nade such progress in the last decade.
1 reply 0 retweets 4 likesShow this thread -
When
@steveklabnik talked about c10k he didn’t mention the 10k lines of C it took to write it then. Maybe thats an exaggeration but not sure.1 reply 0 retweets 3 likesShow this thread -
I mean Nginx was the only comprehensive implementation of c10k for a long time and I think its over 10k loc and a few library calls in the performance serious modern languages are getting close.
3 replies 0 retweets 1 likeShow this thread -
Replying to @justincormack
Apache 2.0 was c100k in 2002. I have the receipts!
1 reply 1 retweet 4 likes -
Replying to @colmmacc
Yeah
@steveklabnik did mention that it changed its design pretty soon (I still remember the mass switch to Nginx and the less long lasting lighttpd). I would love to see a graph of the increase. There was so much innovation in a short period.2 replies 0 retweets 0 likes -
Replying to @justincormack @steveklabnik
Oh god this took me down a rabbit hole. http://www.stdlib.net/~colmmacc/Apachecon-EU2005/scaling-apache-handout.pdf … , we could do c100k, but c27k was our actual record in production. Almost all of it was scheduler and VM tuning though ...pic.twitter.com/3j2bfSCqKO
2 replies 1 retweet 6 likes -
I forgot too that we did profiling to find the optimal buffer sizes with the best mechanical sympathy. Around 30KiB at the time. I wonder what they are now.pic.twitter.com/yZ91y7eLLG
1 reply 0 retweets 4 likes -
Replying to @colmmacc @steveklabnik
“Linux’s sendfile implementation does not operate on our hardware (or any that we can find) without corrupting IPv6 sessions” ha goid old days.
2 replies 0 retweets 1 like
I think later @dwmal1 pinned this down to faulty Intel TOE.
-
-
I've forgotten exact details, but remember that FreeBSD had different offload for IPv4/IPv6, so some NICs only enabled one. Sendfile was also quite different FreeBSD, but on both platforms would result in larger chunks of data being handed off, increasing the chance of offload.
0 replies 0 retweets 1 likeThanks. Twitter will use this to make your timeline better. UndoUndo
-
Loading seems to be taking a while.
Twitter may be over capacity or experiencing a momentary hiccup. Try again or visit Twitter Status for more information.