mongoDB mapreduce 花了很长时间运行 3m 文档

Question

我有一个包含 300 万个文档的集合。每个文档有 40 个字段。字段如下所示。

{
    "b_date" : "2016-04-05",
    "d_date" : "2016-06-25",
    "pos" : "MISC",
    "origin" : "DXB",
    "destination" : "HGA",
    "pax" : 1,
    "pax_1" : 2
 },   
{
    "b_date" : "2016-04-05",
    "d_date" : "2016-06-25",
    "pos" : "MISC",
    "origin" : "DXB",
    "destination" : "HGA",
    "pax" : 4,
    "pax_1" : 5
 },   
{
    "b_date" : "2016-04-05",
    "d_date" : "2016-06-26",
    "pos" : "MISC",
    "origin" : "DXB",
    "destination" : "HGA",
    "pax" : 3,
    "pax_1" : 3
 }

现在我想通过分组 b_date、d_date、pos、origin 来得到 pax 和 pax_1 的总和， destination 个字段。 cumulative pax 是 pos、origin、destination 字段的分组，但 cumulative pax 和 pax_1 应根据 b_date 和 d_date 的升序增加.

Expected result is.

{
    "_id.dep_date" : "2016-04-05",
    "_id.sale_date" : "2016-06-25",
    "_id.pos" : "MISC",
    "_id.origin" : "DXB",
    "_id.destination" : "HGA",
    "value.pax" : 5,
    "value.cumulative_pax":5,
    "value.pax_1" : 7,
    "value.cumulative_pax_1":7,

 },   
{
    "_id.dep_date" : "2016-04-05",
    "_id.sale_date" : "2016-06-26",
    "_id.pos" : "MISC",
    "_id.origin" : "DXB",
    "_id.destination" : "HGA",
    "value.pax" : 3,
    "value.cumulative_pax":8,
    "value.pax_1" : 3,
    "value.cumulative_pax_1":10,
 }

我的 mapReduce 代码

db.collection.mapReduce(
function() {
    emit(
    {
    "pos" : this.pos,
    "origin" : this.origin,
    "destination" : this.destination,
    'dep_date': this.d_date,
    'sale_date': this.b_date,

},
    {
        'pax':this.pax,
        'pax_1':this.pax_1,
    }
    );
}
,
function(key,values) {
    paxt = 0;
    paxt_1 = 0;
    for (var i in values){
    paxt += values[i].pax;
    paxt_1 += values[i].pax_1;
    }
    return {'pax':paxt,
    'pax_1':paxt_1,
    };
}
,
{
    'scope':{
        'pos':'',
        'origin':'',
        'destination':'',
        'dep_date': '',
        'sale_date': '',
        'result':{}
    }
    ,
    'finalize':function(key,value) {
        if (pos != key.pos || 
            origin != key.origin || 
            destination != key.destination || 
            ){  
            result['pax'] = 0;
            result['pax_1'] = 0;
            result['cumulative_pax'] = 0;
            result['cumulative_pax_1'] = 0;
        }
            result['pax'] += value.pax;
            result['cumulative_pax'] = value.pax;
            result['pax_1'] += value.pax_1;
            result['cumulative_pax_1'] = value.pax_1;
            pos = key.pos;
            origin = key.origin;
            destination = key.destination;
            dep_date = key.dep_date;
            sale_date = key.sale_date;

        return result;
        }
    ,
    'out':'some_collection'
    }
)

这张地图减少了返回的预期值，但它花了很多时间，比如 3 个小时。是因为 'b_date' 和 'd_date' 是字符串格式的日期吗？或者如何优化。
聚合在 3 分钟内返回结果，但我无法通过使用聚合获得累积 pax。

Answer 1

Map Reduce代码，

db.collection.mapReduce(
function() {
    emit(
    {
    "pos" : this.pos,
    "origin" : this.origin,
    "destination" : this.destination,
    'dep_date': this.d_date,
    'sale_date': this.b_date,

},
    {
        'pax':this.pax,
        'pax_1':this.pax_1,
    }
    );
}
,
function(key,values) {
    paxt = 0;
    paxt_1 = 0;
    for (var i in values){
    paxt += values[i].pax;
    paxt_1 += values[i].pax_1;
    }
    return {'pax':paxt,
    'pax_1':paxt_1,
    };
}
,
{
    'scope':{
        'pos':'',
        'origin':'',
        'destination':'',
        'dep_date': '',
        'sale_date': '',
        'result':{}
    }
    ,
    'finalize':function(key,value) {
        if (pos != key.pos || 
            origin != key.origin || 
            destination != key.destination || 
            ){  
            result['pax'] = 0;
            result['pax_1'] = 0;
            result['cumulative_pax'] = 0;
            result['cumulative_pax_1'] = 0;
        }
            result['pax'] += value.pax;
            result['cumulative_pax'] = value.pax;
            result['pax_1'] += value.pax_1;
            result['cumulative_pax_1'] = value.pax_1;
            pos = key.pos;
            origin = key.origin;
            destination = key.destination;
            dep_date = key.dep_date;
            sale_date = key.sale_date;

        return result;
        }
    ,
    'out':'some_collection'
    }
)

mongoDB mapreduce 花了很长时间运行 3m 文档

mongoDB mapreduce taking so long time for running 3m document

aggregate

mapreduce

mongodb

mongoDB mapreduce 花了很长时间 运行 3m 文档

mongoDB mapreduce taking so long time for running 3m document

aggregate

mapreduce

mongodb

mongoDB mapreduce 花了很长时间运行 3m 文档