使用PHP与Map和Reduce MongoDB的经典平均演算

Question

I need to calculate the averages for a chart.

I have 15k rows in the database, my index is the time.

I did it in two different way:

1) I repeat on the interval of time (for each interval) : - raw data request between the dates - average calculation in PHP for this interval

2) Map and Reduce: for each interval the reduce function is counting the data, then in the finalize function I make the average.

    m = function() { 
        var k = new Date(this.date);
        k.setSeconds(0);
        k.setMilliseconds(0);
        emit(
            k, { 
                note: this.note
            }
        );
    }
    r = function(key, values) {
        var reduced = { 
                note:0,
                count:0,
                noteAvg:0,
                };
        values.forEach(function(val) {
            reduced.note += val.note; 
            reduced.count += val.count;
        });
        return reduced;
    }
    f = function(key, reduced) {
        reduced.noteAvg = reduced.note / reduced.count;
        return reduced;
    }

    $data_graph = $this->db->command(array(
        "mapreduce" => "notes",
        "map" => $map,
        "reduce" => $reduce,
        "finalize" => $finalize,
        "query" => $req,
        "out" => array("inline"=>1)
    ));

The second solution is a lot of time slower than the first. Why? Should I try to use more data to compare?

I tried on MongoLab (free version) and with my local mongo server and nothing change.

Thanks :)

Answer 1

It sounds like you're mapping all data and filtering it out with reduce rather than restricting the query to the same subset your PHP query is getting.

If you are not already doing so, add a {query:{}} parameter to your mapreduce call as documented here.

This will only pass the subset of documents satisfying the query to the map/reduce operation.