Academy

MongoDB Reporting and Analytics with icCube

2nd May 2014

What is NoSQL Reporting?

NoSQL Databases are great for high scalability and flexible data storage but reporting can be tricky because of the lack of defined structure and the performance cost. Multi-Dimensional Reporting tools are great for light speed analytics and reporting on structured data but do not work well on fast changing unstructured data. Why not combine the best of both worlds? MongoDB Reporting with icCube allows you to create advanced reports with NoSQL databases.

MongoDB Reporting with icCube

As a new feature icCube integrates natively MongoDB: Map/Reduce, Aggregation and complex JSON Object Types are now features fully available in all icCube versions (including free Community).

MongoDB Reporting with icCube brings fast real time in-memory data analysis to MongoDB and in a near future to other NoSQL databases. Once the data has been imported into icCube, you can use the full power of MDX+ for the most demanding data analysis. You will be able to report with your favorite XMLA reporting tool (e.g. Excel) as well with icCube’s Reporting.

Note that this MongoDB data can be combined with data coming from other data sources (e.g. RDBM, Excel, CSV, BigQuery amongst others) to create more complex and powerful analysis.

Let’s imagine we’ve the following collection/documents in MongoDB:

{
    customer: {
        id: "2342WKO"
        gender: “male”
    },
    shoppingCart: {
        id: "423KIL"
        date :”2012-02-23T12:23:00.000Z”
    },
    products: [{
        sku: "2334FGH",
        quantity: 1
    }, {
        sku: "8764KJU",
        quantity: 2
    }]
    ...
 }

How to integrate into icCube ?

First we have to bring the data into the icCube server. For this we’ve two different ways of creating queries: aggregation pipelines vs map/reduce commands. Both will return a list of documents (JSON records).

When importing these documents icCube will automatically transform the JSON records into structures for optimal many to many modeling. There is no need to denormalize, unwind, each JSON record, for example generate multiple lines for each product.

This new features will keep minimum data on the wire, reduce memory footprint as well as improve overall MDX+ query performance.

Once done you’ll have to create your dimensions and measures matching your business attributes and you’d be ready for data analysis and reporting.

Last Words

This is the end of this quick overview of MongoDB support in icCube. We invite you to experiment with MongoDB by downloading icCube. For more questions and details please use our forum.

By David Alvarez