MongoDB 2.6 is $out

Last modified:

Imagine having a tool that can automatically detect JPA and Hibernate performance issues. Wouldn’t that be just awesome?

Well, Hypersistence Optimizer is that tool! And it works with Spring Boot, Spring Framework, Jakarta EE, Java EE, Quarkus, or Play Framework.

So, enjoy spending your time on the things you love rather than fixing performance issues in your production system on a Saturday night!

Introduction

MongoDB is evolving rapidly. The 2.2 version introduced the aggregation framework as an alternative to the Map-Reduce query model. Generating aggregated reports is a recurrent requirement for enterprise systems and MongoDB shines in this regard. If you’re new to it you might want to check this aggregation framework introduction or the performance tuning and the data modelling guides.

Let’s reuse the data model I first introduced while demonstrating the blazing fast MongoDB insert capabilities:

{
    "_id" : ObjectId("5298a5a03b3f4220588fe57c"),
    "created_on" : ISODate("2012-04-22T01:09:53Z"),
    "value" : 0.1647851116706831
}

MongoDB 2.6 Aggregation enhancements

In the 2.4 version, if I run the following aggregation query:

db.randomData.aggregate( [
{
    $match: {
        "created_on" : {
            $gte : new Date(Date.UTC(2012, 0, 1)),
            $lte : new Date(Date.UTC(2012, 0, 10))
        }
    }
},
{
    $group: {
        _id : {
            "minute" : {
                $minute : "$created_on"
            }
        },
        "values": {
            $addToSet: "$value"
        }
    }
}]);

I hit the 16MB aggregation result limitation:

{
    "errmsg" : "exception: aggregation result exceeds maximum document size (16MB)",
    "code" : 16389,
    "ok" : 0
}

MongoDB documents are limited to 16MB, and prior to the 2.6 version, the aggregation result was a BSON document. The 2.6 version replaced it with a cursor instead.

Running the same query on 2.6 yields the following result:

db.randomData.aggregate( [
{
    $match: {
        "created_on" : {
            $gte : new Date(Date.UTC(2012, 0, 1)),
            $lte : new Date(Date.UTC(2012, 0, 10))
        }
    }
},
{
    $group: {
        _id : {
            "minute" : {
                $minute : "$created_on"
            }
        },
        "values": {
            $addToSet: "$value"
        }
    }
}])
.objsLeftInBatch();

I used the cursor-based objsLeftInBatch method to test the aggregation result type and the 16MB limitation no longer applies to the overall result. The cursor inner results are regular BSON documents, hence, they are still limited to 16MB, but this is way more manageable than the previous overall result limit.

The 2.6 version also addresses the aggregation memory restrictions. A full collection scan such as:

db.randomData.aggregate( [
{
    $group: {
        _id : {
            "minute" : {
                $minute : "$created_on"
            }
        },
        "values": {
            $addToSet: "$value"
        }
    }
}])
.objsLeftInBatch();

can end up with the following error:

{
    "errmsg" : "exception: Exceeded memory limit for $group, but didn't allow external sort. Pass allowDiskUse:true to opt in.",
    "code" : 16945,
    "ok" : 0
}

So, we can now perform large sort operations using the allowDiskUse parameter:

db.randomData.aggregate( [
{
    $group: {
        _id : {
            "minute" : {
                $minute : "$created_on"
            }
        },
        "values": {
            $addToSet: "$value"
        }
    }
}]
,
{
    allowDiskUse : true
})
.objsLeftInBatch();

The 2.6 version allows us to save the aggregation result to a different collection using the newly added $out stage.

db.randomData.aggregate( [
{
    $match: {
        "created_on" : {
            $gte : new Date(Date.UTC(2012, 0, 1)),
            $lte : new Date(Date.UTC(2012, 0, 10))
        }
    }
},
{
    $group: {
        _id : {
            "minute" : {
                $minute : "$created_on"
            }
        },
        "values": {
            $addToSet: "$value"
        }
    }
},
{
    $out : "randomAggregates"
}
]);
db.randomAggregates.count();

New operators have been added such as let, map, cond, to name a few.

The next example will append AM or PM to the time info of each specific event entry.

var dataSet = db.randomData.aggregate( [
{
    $match: {
        "created_on" : {
            $gte : new Date(Date.UTC(2012, 0, 1)),
            $lte : new Date(Date.UTC(2012, 0, 2))
        }
    }
},
{
    $project: {
        "clock" : {
            $let: {
                vars: {
                    "hour": {
                        $substr: ["$created_on", 11, -1]
                    },
                    "am_pm": { $cond: { if: { $lt: [ {$hour : "$created_on" }, 12 ] } , then: 'AM',else: 'PM'} }
                },
                in: { $concat: [ "$$hour", " ", "$$am_pm"] }
            }
        }
    }
},
{
    $limit : 10
}
]);
dataSet.forEach(function(document)  {
    printjson(document);
});

Resulting in:

"clock" : "16:07:14 PM"
"clock" : "22:14:42 PM"
"clock" : "21:46:12 PM"
"clock" : "03:35:00 AM"
"clock" : "04:14:20 AM"
"clock" : "03:41:39 AM"
"clock" : "17:08:35 PM"
"clock" : "18:44:02 PM"
"clock" : "19:36:07 PM"
"clock" : "07:37:55 AM"

Conclusion

MongoDB 2.6 version comes with a lot of other enhancements such as bulk operations or index intersection. MongoDB is constantly evolving, offering a viable alternative for document-based storage.