nedb/benchmarks/insert.js

var Datastore = require('../lib/datastore')
  , benchDb = 'workspace/insert.bench.db'
  , async = require('async')
  , commonUtilities = require('./commonUtilities')
  , execTime = require('exec-time')
  , profiler = new execTime('INSERT BENCH')
  , d
  , program = require('commander')
  , n
  ;

program
  .option('-n --number [number]', 'Size of the collection to test on', parseInt)
  .option('-i --with-index', 'Use an index')
  .option('-p --with-pipeline', 'Use pipelining')
  .parse(process.argv);

n = program.number || 10000;

console.log("----------------------------");
console.log("Test with " + n + " documents");
console.log(program.withIndex ? "Use an index" : "Don't use an index");
console.log(program.withPipeline ? "Use an pipelining" : "Don't use pipelining");
console.log("----------------------------");

d = new Datastore({ filename: benchDb, pipeline: program.withPipeline });

async.waterfall([
  async.apply(commonUtilities.prepareDb, benchDb)
, function (cb) {
    d.loadDatabase(function (err) {
      if (err) { return cb(err); }
      if (program.withIndex) {
        d.ensureIndex({ fieldName: 'docNumber' });
        n = 2 * n;   // We will actually insert twice as many documents
                     // because the index is slower when the collection is already
                     // big. So the result given by the algorithm will be a bit worse than
                     // actual performance
      }
      cb();
    });
  }
, function (cb) { profiler.beginProfiling(); return cb(); }
, async.apply(commonUtilities.insertDocs, d, n, profiler)
], function (err) {
  profiler.step("Benchmark finished");

  if (err) { return console.log("An error was encountered: ", err); }
});