sandshark

• zero dependencies!
• aiming for 100% test coverage, currently @ ~70%
• express.js syntax like with raw http performance
• setup & middlewares should work right out of the box with your expressjs project
• this is still "work in progress"!

currently in TODO

• improving performance for inc. connections (not better than expressjs on benchmark results with keep-alive missing)
• implement more express apis, a lot of stuff is still missing e.g. res.type(), app.disable()
• implement custom cookie, query & body parser that can be toggled via config
• microbench to reach 99,5% performance of raw http module
• improve test coverage to 100%

run example

• via npm start

run tests

• via npm test

run benchmarks

npm run benchmark sandshark
npm run benchmark express
npm run benchmark raw
npm run benchmark restify
npm run benchmark hapi

• or run all benchmarks in comparison via npm run benchmarks or npm run benchmark all

• to compare expressjs and sandshark run npm run compare

• each benchmark will spawn a server and worker making requests in their own child processes

• the loadtest worker is limited to a single thread and makes at least 2500 requests

• the benchmarks currently lack additional features such as body parsing, query params, path variables and cookies

• there is a good chance to have far more place for optimizations

• since hapi feels rather slow, feel free to send a pull requests if you think that I got something wrong

benchmark results

• on a MacBook Air 11' 2014 Intel i5 quad @ 1.4 GHz
• node version 6.5.0 & server modules are present in package.json devDependencies

[
  { raw: 1725 },
  { sandshark: 1692 },
  { express: 1223 },
  { restify: 1084 },
  { hapi: 687 }
]

• with keep-alive turned off

[
  { express: 488 },
  { raw: 426 },
  { sandshark: 373 },
  { restify: 318 },
  { hapi: 314 }
]

disclaimer

• i started this project after reading Node Performance hapi vs express vs restify
• since their benchmarks are from March 2015, i guess they are outdated ;)
• in their benchmarks raw http performance hits 8500 req/s and express only 3000 req/s
• my benchmarks actually show expressjs way closer to raw http performance so another guess is that their benchmarks contain possible errors, sadly they have not posted them online

furthermore

• i would love to have further discussions on this, so please feel free to open issues or hit me up on twitter @silentleave or my website