Well spectators, too much is never enough when it comes to understanding the power of cubes. Bear with me here. If you've spoken to us at all the word cube will have floated into conversation. If we've chatted a lot then its probably come up quite a bit.
So what power does this technology have over us that we evangelise so readily? First some background and history. Cubes are an analytical component - a way of storing data and pivoting it to allow for analysis across a broad range of perspectives at high speed. Wikipedia tells us that it has 3 essential analytical operations, aggregation, drilling and slice and dice. The technology that produces cubes is called OLAP (Online Analytical Processing - a term coined by the venerable Edgar F Codd in 1993 with his "A Relational Model of Data for Large Shared Data Banks").
The key to getting why cubes are about the best slicer invented since bread is in the name - Online (connotes speed and performance) and Analytical (speaks to a technology specifically designed for Analysis). In practice you find OLAP to be less a technology than a way of thinking or comprehending the multidimensional nature of data and its potential.
The alternative/twin/antithesis to OLAP is OLTP query and reporting tools. Online Transactional Processing, (for this discussion I'll consider relational databases as synonymous with OLTP). With OLTP you have the classic relational table structure of data, tables joined by keys to other tables storing vast amounts of structured data.
Now there are other types of data in the world, unstructured, Big etc.. but for this discussion, the vast majority of data used effectively in business today, through ERP's and related systems, arrives as OLTP or relational. At that point, the would-be analyst can decide to use relational reporting and query tools to report from that data or OLAP.
That decision requires consideration of advantages and efficacy. Now for a long time, we over here on the OLAP side of the world, having had the epiphany of multidimensional thinking when it comes to data, would look askance at our friends with their relational reporting tools and say, "good lord, you have to wait how long for a query to run?". OLAP has blistering analytical query response you see. Game over, think no more about. There were other big advantages of OLAP but performance was such a radical stand out that these did not need to be resorted to.
In time, OLTP and relational have continued to improve performance with in-memory processing, column indexing etc - they’re getting quick now. So now we have to look at the other reasons why OLAP is the ultimate gateway technology for business people who want to rapidly collect, interpret and act on their data, both singly and in collaboration.
Don’t get me wrong, OLAP performance is still phenomenal, but what else is there? There's meta-data magic. Meta data is data about data. Meta data gives context. So I may have a number, say 50,000. But until I know that 50,000 represents the number of Salmon (meta data point 1), caught in Yellowstone National Park (meta data point 2) by Yogi Bear (meta data point 3), on a given Sunday (meta data point 4) and it was raining (meta data point 5), I don’t know much about the 50,000. Now when those meta data points live in OLAP dimensions and hierarchies so I know that Yellowstone is in a particular grouping of Parks that add up to USA Parks or that my given Sunday falls in a month, a season and a year, and that Yogi is the alpha in a pack which is one of several packs, I start to get enormous analytical power to compare, drill and understand how my data fits to together and what it can tell me.
Some of the immense power of the cube comes from the fact that (with the right cube viewer, say CALUMO) I can come to a cube knowing very little, but the meta data that is baked into its dimensions, hierarchies and measures tell me so much about the information I am engaging with. If I simply look at a cube of a given business I can tell immediately how its departments organize, its company structure, its products, its sales regions and the list goes on. And I can do that entirely myself - without ever having to be given a link to a table or build a pivot table or asking the quiet older gentleman in the corner exactly how this business fits together. As an analyst, I build a cube once and share it and my knowledge is instantly leveraged for literally years to come for any of those whom I serve. Through the way we do reporting, CALUMO extends this inherent magic in OLAP to a full range of reporting activities in your business.
Many of our friends over in the visualisation/dashboarding space say that "cubes are too hard". That sadly is misdirection. Cubes aren’t hard. Sometimes getting everyone to agree on the meta data can be challenging, but the rewards are immense. Others suggest that what you need is to just mash your data together when you want to get insights. The trouble with mashing is that you often just get a mash. Some insights perhaps but the unreliable means of their creation makes the insight difficult to action. You get many different mashes purportedly for the same insight - and the insights look different. One version of the truth gets some serious damage along the way and splinters into many.
We have recently extended CALUMO’s capability to provide interactive visualisation of data. We call it Visual Data Discovery – and it’s basically interactive visualisations for cubes. Watching this through the R&D phase to delivery has been such a delight because it means that even more of the power of OLAP is exposed. Navigating cubes with charts is delightful. And you know that the data you are moving through is the reliable, one version of the truth and, because of the meta data magic of the cube, completely in context and repeatable.
Thought you knew enough about cubes already? There's always more to know. I haven't even started on the power that cubes bring to modeling, budgeting and planning - lets save that for another time.