Talk about MongoDB

about how I screwed up in the past (and what I've learned)

At the end of this talk, ask me in English, Spanish or Portuguese.

This Slides are open source http://github.com/crodas/Slides

a database (like MySQL)

without SQL

friendly as Doctrine (but without overhead)

MyDoctrineWithoutOverhead

bad joke I know

db.users.save({
    _id: 4290392,
    name: "César Rodas",
    emails: [
        {email: "saddor@gmail.com", jabber:true, mail:true},
        {email: "crodas@php.net", mail:true, jabber: false},
        {email: "jabber@crodas.org", mail:false, jabber: true},
    ],
    likes: ["PHP", "NodeJS", "MongoDB"],
    age: 23,
    address: {
        country: "Paraguay",
        city: "Asunción",
    }
});

• Semi-structured model
• Array (PHP)
• Dictionary (Python)
• Hash (Perl)
• Object (without methods)

• Everything is a document
• Fast at query time.
• Simplest data distribution.
• Less painful migrations.
• UTF-8 friendly.
• Type-rich supports dates, regular expressions, code, binary data, and more.

var cursor = db.users.find({"age": 4290392});

cursor = db.users.find({"likes": "PHP", "age":23});

cursor = db.users.find({"address.country": "Paraguay"});

cursor = db.users.find({"age": {"$gt": 18}});

cursor = db.users.find({"emails.mail": true});

cursor = db.users.find({"emails": {"$elemMatch":{"mail": false, "jabber":true}}});

• Indexes are optional.
• Speed up queries, slow inserts (updates?).
• Could be single-key, compound, unique, non-unique, and geospatial.
• The query optimizer will try a number of different query plans when a query is run and select the fastest, periodically resampling.

db.users.find({sex:"male"});
// Takes 341ms
        

db.users.find({sex:"male"}).explain()
{
        "cursor" : "BasicCursor",
        "nscanned" : 11732,
        "nscannedObjects" : 11732,
        "n" : 843,
        "millis" : 36,
        "nYields" : 0,
        "nChunkSkips" : 0,
        "isMultiKey" : false,
        "indexOnly" : false,
        "indexBounds" : {

        }
}

db.users.ensureIndex({sex:"male"})
db.users.find({sex:"male"}).sort({age:-1}).explain()
{
        "cursor" : "BtreeCursor sex",
        "nscanned" : 843,
        "nscannedObjects" : 843,
        "n" : 843,
        "scanAndOrder" : true,
        "millis" : 27,
        "nYields" : 0,
        "nChunkSkips" : 0,
        "isMultiKey" : false,
        "indexOnly" : false,
        "indexBounds" : {
                "sex" : [
                    ...
                ]
        }
}

db.users.ensureIndex({sex:"male", age:-1})
db.users.find({sex:"male"}).sort({age:-1}).explain()
{
        "cursor" : "BtreeCursor sex__age_-1",
        "nscanned" : 843,
        "nscannedObjects" : 843,
        "n" : 843,
        "millis" : 2,
        "nYields" : 0,
        "nChunkSkips" : 0,
        "isMultiKey" : false,
        "indexOnly" : false,
        "indexBounds" : {
                "sex" : [
                    ...
                ],
                "age" : [
                    ...
                ]
        }
}

db.foo.ensureIndex({"address.country":1});
    

address: {
    country: "Brazil",
    city: "Sao Paulo",
}

address: [
    {
        country: "Brazil",
        city: "Sao Paulo",
    },
    {
        country: "Asunción",
        city: "Paraguay",
    },
]

db.foo.ensureIndex({"address.country":1});
db.foo.ensureIndex({"address":1});
    

• Measures slow operations
• It is saved at system.profile collection
• Measure query, update, insert and getmore
• Similar to explain()
• http://www.mongodb.org/display/DOCS/Database+Profiler

• Data redundancy
• Embedding vs Referencing