node.js - Why are MongoDB upserts so much slower than inserts (with a unique index)? -

i've been testing limits of mongodb see whether work upcoming project , i've noticed upserts quite slow compared inserts.

of course, i'd expect them slower, not (almost) order of magnitude slower (7400 vs 55000 ops/sec). here's (nodejs native driver) bench-marking code used:

(async function() {    let db = await require('mongodb').mongoclient.connect('mongodb://localhost:27017/mongo-benchmark-8764824692947');   db.collection('text').createindex({text:1},{unique:true})    let batch = db.collection('text').initializeorderedbulkop();   let totalopcount = 0;   let batchopcount = 0;   let start = date.now();   while(1) {      totalopcount++;     batchopcount++;     if(batchopcount === 1000) { // batch 1000 ops @ time       await batch.execute();       batch = db.collection('text').initializeorderedbulkop();       batchopcount = 0;       let secondselapsed = (date.now() - start)/1000;       console.log(`(${math.round(totalopcount/secondselapsed)} ops per sec) (${totalopcount} total ops)`)     }      /////////  insert test  ///////// (~55000 ops/sec)     // batch.insert({text:totalopcount});      /////////  upsert test  ///////// (~7400 ops/sec)     let text = math.floor(math.random()*1000000);     batch.find({text}).upsert().updateone({$setoninsert:{text}});      if(totalopcount > 500000) {       console.log("<< finished >>");       await db.dropcollection('text');       db.close();       break;     }    }  })(); 

you can run pasting index.js, running npm init -y , npm install --save mongodb , node .

when upsert document, mongo engine has check whether there's existing document matches it. might have explained of slowdown, doesn't insert command on unique index require same collision checking? thanks!

edit: turns out $setoninsert is needed else duplicate key errors.

i made each batch bigger changing if(batchopcount === 1000) if(batchopcount === 50000), , got ~90000 ops/sec insert , ~35000 ops/sec upsert. i'm not sure why batch size makes relative difference. makes sense smaller batches result in fewer ops/sec (communication overhead), i'm not sure why upserts suffer more inserts, though topic separate question.

this (roughly) 3-fold speed difference closer performance difference between 2 operations i'd expect, still seems upserts little slower should be, given inserts must check collisions (due unique index). half-answer, i'm going leave question open in hope mongo pro come along , provide more complete answer.