Computer Science through Streams
summary
Demonstrates that Node streams (fs.createReadStream, piping, gzip, a CSV transform and http.createServer) let a server send a 10 GB file and a query of 65,000 rows over a slow airplane/EC2-micro link by using TCP-style high water mark flow control rather than buffering everything. The default 16KB buffer and backpressure make streaming use about five times less heap than a buffered approach; the implementation works slowly but reliably. Tools referenced: libuv, streamviz, stream handbook and nodeschool exercises.
AI-written summary · about
chapters (7)
- 0:00Start
- 4:28BACKPRESSURE
- 8:05WHAT HAPPENS IF A STREAM IS OUT POUR CONTROL?
- 9:12having a mechanism for flow control is ESSENTIAL in an environment where machines of DIVERSE NETWORK SPEEDS connunicate.
- 15:01STREAM SUCCESS BUFFER FAIL
- 16:42DIVERSE NETWORK IS for SPEEDS
- 16:53WHY? ALL the THINGS ARE STREAMS
related talks
28mThe good, the bad and the ugly of nodejs streams
34mUnderstanding Streams, Signals and Events in Nodejs
33mTaking Node.js to The Ex-Stream: Stream to Scale
34mYou Don't Know Node.js
49mNode JS
33mStream into the Future
25mA JavaScript Parody: A Fun Journey through Hot Topics of Node.js and JS
23mStreaming Algorithms in Javascript and Node.js