Notes from #lwshttp

Playlister - Developing a new music product at the BBC - Sara Gonzalo

  • Why? People searching for ‘that some that was on last week’
  • @bbcplaylister
  • So many track IDs, lets use our own
  • Developing the new API using the docs from the old one
  • Nobody knows what ‘export’ means, they just want to ‘listen’

What to expect from HTTP/2 - Mark Nottingham

  • Performance = Attention = £££
  • There are a lot of techniques people use today to get the most out of http performance - today
    • Sprites
    • Sharding (multiple hosts)
    • Concatenation
  • All these techniques are hacks
  • Recurring theme is “Eliminate Requests”
  • Why are HTTP requests so expensive?
    • HTTP/1 uses TCP poorly
    • HTTP Headers are quite verbose - a lot of duplication between header requests
  • What HTTP/2?
    • It started with SPDY (speedy)
    • Separate the semantics of the protocol from the framing
    • Full multiplexing of headers and data, back and forth, within a stream
    • Header Compression
    • Server Push - optimistically sending what data the server thinks the client will need next (CSS straight after HTML)
  • Spec should be done by the end of the year
  • Working draft of HTTP/2 is implemented on the twitter servers now
  • HTTP semantics won’t change… But, leaked abstractions will.
  • Rethink things
    • Load balancing
    • Fallovers
  • Use TLS. Firefox and Chrome will require it
  • Requests will be cheaper
  • Getting from 1 to 2
    • https: ALPN
    • http: Inband upgrade ‘dance’, hint that HTTP/2 is available on another port
  • http://http2.github.io
  • @http_2