MongoDB Primary Keys Are Your Friend

All documents in a MongoDB collection have a primary key dubbed _id. This field is automatically assigned to a document upon insert, so there’s rarely a need to provide it. What’s interesting about the _id field is that it is time based. That is, the underlying type of _id, which is ObjectId, is a 12-byte BSON type, and 4 of those bytes represent the seconds since Unix epoch.

What’s also special about the _id field is that it is automatically indexed as you can see below by calling getIndexes on any collection.

All MongoDB collections have an _id field as an index:
 

> db.things.getIndexes()
[
     {
          "v" : 1,
          "key" : {
               "_id" : 1
          },
          "ns" : "test.things",
          "name" : "_id_"
     }
]

And as everyone remembers from traditional RDBMSs, indexes are important because they can make document retrieval faster; nevertheless, indexes do consume memory and there is a slight performance penalty when inserting documents as all corresponding indexes must be updated. Thus, while you should seriously consider using indexes, you need to be economical in their usage.

Naturally, searching by a document’s _id is only convenient when you know it. More often than not, documents are searched via other fields and if you find yourself searching via a time series, such as created_at then you are in for a treat.

Imagine a collection dubbed logs that contains simple documents capturing various log messages. A sample document could look like so:

A simple document in a logs collection:

{
     "_id" : ObjectId("51c4ab6d4d6906d494460728"),
     "message" : "crashed, no such method exception",
     "type" : "crash",
     "created_at" : ISODate("2013-06-21T19:37:17.992Z")
}

What if I wanted to find all log messages for some date, like today? I could write my query like so:

Finding all logs created since June 20th, 2013:

db.logs.find({created_at:{'$gt': new Date(2013, 5, 20)}})

If I throw an explain to that query, I can see that because I do not have an index on created_at, a basic cursor is leveraged and all documents in the collection were scanned in order to retrieve my result.

An explain plan attached to my find:

> db.logs.find({created_at:{'$gt': new Date(2013, 5, 20)}}).explain()
{
     "cursor" : "BasicCursor",
     "isMultiKey" : false,
     "n" : 2,
     "nscannedObjects" : 4,
     "nscanned" : 4,
     "nscannedObjectsAllPlans" : 4,
     "nscannedAllPlans" : 4,
     "scanAndOrder" : false,
     "indexOnly" : false,
     "nYields" : 0,
     "nChunkSkips" : 0,
     "millis" : 0,
     "indexBounds" : {

     },
     "server" : "ghome-computer.home:27017"
}

As you can see, searching via the created_at field can be inefficient; thus, you might be tempted to throw an index on that field. This would naturally make that particular query more efficient, however, you would incur the cost of a new index which is more memory consumed and inserts would be slightly slower due to an update to that newly created index.

As it turns out, because the _id field embeds Unix epoch in it, you can just as easily craft a find expression without including the created_at field. For example, the MongoDB Ruby driver allows you to create ObjectId’s from a Time like so:

Creating a new ObjectId via the from_time factory method:

yesterday = Time.now - (60*60*(24*1))
custom_id = BSON::ObjectId.from_time(yesterday)
=> BSON::ObjectId('51c397800000000000000000')

As you can see, I’ve created a new ObjectId via the from_time factory method. 51c397800000000000000000 is a hexadecimal representation and the first 8 digits represent the time with everything else zeroed out.

Now I can leverage my custom_id in any find expression. Via the Ruby driver, I can also attach an explain, which’ll demonstrate the usage of the free _id index.

Using a date derived ObjectId forces a find to use the _id index:

mongodb[:logs].find({_id: {'$gt' => custom_id}}).explain

=> {"cursor"=>"BtreeCursor _id_", "isMultiKey"=>false, "n"=>1, "nscannedObjects"=>1, "nscanned"=>1, ....}

If you see BtreeCusor, then you know you’re using an index; if you see BasicCursor, you know you’re not.

Thus, if you find yourself executing queries and creating indexes for some time or date field like created_at, you might be better off just using Mongo’s _id field as it already embeds the notion of created at and is indexed by default. Dig it?
 

Reference: MongoDB Primary Keys Are Your Friend from our JCG partner Andrew Glover at the The Disco Blog blog.

Do you want to know how to develop your skillset to become a Java Rockstar?

Subscribe to our newsletter to start Rocking right now!

To get you started we give you two of our best selling eBooks for FREE!

JPA Mini Book

Learn how to leverage the power of JPA in order to create robust and flexible Java applications. With this Mini Book, you will get introduced to JPA and smoothly transition to more advanced concepts.

JVM Troubleshooting Guide

The Java virtual machine is really the foundation of any Java EE platform. Learn how to master it with this advanced guide!

Given email address is already subscribed, thank you!
Oops. Something went wrong. Please try again later.
Please provide a valid email address.
Thank you, your sign-up request was successful! Please check your e-mail inbox.
Please complete the CAPTCHA.
Please fill in the required fields.

One Response to "MongoDB Primary Keys Are Your Friend"

  1. staalla says:

    Its a nice trick although consider this as well, _id should also be unique for the collection (logs). Its certainly easier to read without an extra index but the processes which write the log messages to the collection, using custom id generated based on time(depending on the granularity or clock skew on different systems or load) may not be unique, these messages could be lost with duplicate key exception.

Leave a Reply


× 2 = ten



Java Code Geeks and all content copyright © 2010-2014, Exelixis Media Ltd | Terms of Use | Privacy Policy | Contact
All trademarks and registered trademarks appearing on Java Code Geeks are the property of their respective owners.
Java is a trademark or registered trademark of Oracle Corporation in the United States and other countries.
Java Code Geeks is not connected to Oracle Corporation and is not sponsored by Oracle Corporation.
Do you want to know how to develop your skillset and become a ...
Java Rockstar?

Subscribe to our newsletter to start Rocking right now!

To get you started we give you two of our best selling eBooks for FREE!

Get ready to Rock!
You can download the complementary eBooks using the links below:
Close