MongoDB Field Bucketing Benchmarks

bucket_tests.js

function logMessage(message) {
    var d = new Date()
    print(`${d.getHours()}:${d.getMinutes()}:${d.getSeconds()}.${d.getMilliseconds()}: ${message}`)
}

coll_size = 50000 // Number of documents in the collection we're aggregating
subdoc_size = 2500 // Number of subfields in the columns subdocument
subdocument_targets = [10, 25, 50, 100, 250, 500, 1000, 1500, 2500]

// Non-bucketed tests

logMessage("Dropping bucket collection.")
db.bucket.drop()

// Create the columns subdocument.
columns_subdoc = {}
for (var i = 0; i < subdoc_size; ++i) {
    columns_subdoc[`col${i + 1}`] = 5;
}

logMessage(`Inserting ${coll_size} non-bucketed documents.`)
for (var i = 0; i < coll_size; i++) {
        db.bucket.insert({
            _id: i,
            columns: columns_subdoc
        })
}
logMessage("Finished inserting non-bucketed documents.")

for (var subdoc_target_idx = 0; subdoc_target_idx < subdocument_targets.length; ++subdoc_target_idx) {
    subdoc_target = subdocument_targets[subdoc_target_idx]

    //logMessage(`Using collection with ${coll_size} documents and targeting subfield ${subdoc_target}.`)
    //logMessage("Running non-bucketed benchmark 15 times.")
    
    var non_bucketed_benchmark_total = 0;
    for (var i = 0; i < 15; i++) {
        //logMessage(`Aggregating non-bucketed documents with sum of subfield ${subdoc_target}`)
        var before = new Date()

        db.bucket.aggregate([{
            $group: {_id: null, total: {$sum : `$columns.col${subdoc_target}`}}
        }])

        var after = new Date()
        //logMessage(`Finished aggregating non-bucketed documents with sum of subfield ${subdoc_target}. Took ${after - before}ms`)
        non_bucketed_benchmark_total += (after - before)
    }

    logMessage(`----- RESULTS FOR NON-BUCKETED COLLECTION WITH ${coll_size} DOCUMENTS AND TARGETING SUBFIELD ${subdoc_target} -----`)
    logMessage(`----- Average aggregation time ${non_bucketed_benchmark_total / 15}ms ---------------------`)
    logMessage("-------------------------------------------------------------------------------------------")
}

// Bucketed tests
logMessage("Starting bucketed tests.")

bucket_sizes = [16, 32, 64, 128]

for (var bucket_size_idx = 0; bucket_size_idx < bucket_sizes.length; ++bucket_size_idx) {
    bucket_size = bucket_sizes[bucket_size_idx]
    logMessage(`Using buckets with ${bucket_size} subfields.`)

    logMessage("Dropping bucket collection.")
    db.bucket.drop()

    // Create the bucketed columns subdocument.
    var bucket_idx = 0
    columns_subdoc = {}
    for (var i = 0; i < subdoc_size; ++i) {
        if (i % bucket_size == 0) {
            ++bucket_idx;
            columns_subdoc[`bucket${bucket_idx}`] = {}
        }
        columns_subdoc[`bucket${bucket_idx}`][`col${i + 1}`] = 5;
    }

    logMessage("Inserting bucketed documents")
    for (var i = 0; i < coll_size; i++) {
        db.bucket.insert({ 
            _id: i,
            columns: columns_subdoc
        })
    }
    logMessage("Finished inserting bucketed documents")

    for (var subdoc_target_idx = 0; subdoc_target_idx < subdocument_targets.length; ++subdoc_target_idx) {
        subdoc_target = subdocument_targets[subdoc_target_idx]

        //logMessage("Running bucketed benchmark 15 times.")
        target_bucket = Math.floor(subdoc_target / bucket_size) + 1

        var bucketed_benchmark_total = 0

        for (var i = 0; i < 15; i++) {

            //logMessage(`Aggregating bucketed documents with sum of subfield ${subdoc_target}`)
            var before = new Date()

            db.bucket.aggregate([{
                $group: {_id: null, total: {$sum : `$columns.bucket${target_bucket}.col${subdoc_target}`}}
            }])

            var after = new Date()
            //logMessage(`Finished aggregating bucketed documents with sum of subfield ${subdoc_target} Took ${after - before}ms`)

            bucketed_benchmark_total += (after - before)
        }

        logMessage(`----- RESULTS FOR BUCKETED COLLECTION WITH ${coll_size} DOCUMENTS,\n----- ${bucket_size}-FIELD BUCKETS AND TARGETING SUBFIELD ${subdoc_target}`)
        logMessage(`----- Average aggregation time ${bucketed_benchmark_total / 15}ms`)
        logMessage(`--------------------------------`)

    }
}