When your users like to explore their data, Online Analytical Processing (OLAP) technologies are often the answer. OLAP solutions provide users a web-based, interactive tool to drill and explore trends across time, customer, product, geography or any other dimension in the data.
Selecting the right OLAP engine.
We've found the open source Mondrian engine to be a solid choice for many customers. Embedded in the Pentaho platform, or deployable standalone, this platform does a great job of combining the dimensional database design optimizations (e.g. classic star-schema) with high speed caching to produce interactive data exploration. It also works with the growing number of specialized analytic databases like Actian, InfiniDB, Redshift and Vertica.
Another popular platform we've used is the Microsoft SSAS stack, which uses the same MDX query language as Mondrian. When looking at OLAP engines, it's important to consider the cost benefit of the platforms.
Developing the ideal OLAP schema.
We spend most of our time on OLAP projects getting the core metadata schema right. The metadata layer is the business semantic by which people interact with the data and the mapping between dimensional concepts and physical database design. Its design is critical for both usability and performance. We think of schema development holistically; its about creating intuitive and interactive access to data. Ideal OLAP schema harmonize data and usage.
Developing dimensions and measures that work.
When the data is right, and the database schema is in good shape, developing the dimensions and measures is a whole lot easier. At the same time, one of the most powerful things in OLAP is the fact that you don't need every data element expressed directly in the data. MDX is an incredibly powerful language that can do incredible things with base data, including period over period calculations, regressions, and moving averages . Doing advanced computations and presenting crosstabular results with SQL will make your head spin. This is why we use OLAP technologies.
Amazing analysis with cubes and virtual cubes.
One of the powerful constructs in the Mondrian engine is turning cubes into virtual cubes, allowing incredibly interactive analysis on subjects of data that span various topics. We've done virtual cubes that bring together information in a single interactive OLAP session that users have never been seen before.
Tuned cubes for quick answers.
Mondrian is aggregate aware, which means it's smart enough to send a query to an aggregated table or materialized view. We are expert at aggregate design and Mondrian tuning, providing users incredibly fast results from incredibly detailed data.
Drilling in to see the answers behind the numbers.
When people want to see the details behind the results of a SQL query, this often results in writing another report with another set of parameters. OLAP is built for drill through, allowing you to see the details behind every cell. Powerful stuff.
See what you need, and not what you don't.
Any data framework needs authorization models that only allow people to see the data they need, hiding the data they don't. Mondrian has advanced data authorization levels, including rights at the cube, dimension, hierarchy, and even specific data members.
Seeing OLAP in action.
OLAP engines need a front-end to enable drilling up, down and through the data while also creating powerful data analytic components including dashboards, charts and reports. Paired with Pentaho Analyzer or open source Saiku, Mondrian enables users to interact with cubes via a web-based, wysiwyg interface to create reports and charts that can be saved, shared and embedded in dashboards. Additionally, since Mondrian can act as an XMLA server, other tools and technologies can source data from it, including Excel.
Inquidia gets your data into shape and can deploy OLAP analysis as part of a comprehensive reporting solution for you.