Why It’s Time to Rethink Your Cube-Based BI Flow and Embrace Analytical Views
Are you aware that the OLAP cube is a technology that was introduced way back in the 70’s? It was a key innovation back then, allowing programmers to pull multi-dimensional data from a data warehouse and store it in a format that could be queried easily. In today’s BI and analytics systems, these cubes or extracts are basically pre-aggregated results tables that provide faster application response times, since they don’t need to scan every row of a base fact table when providing the data needed by the application. But because the data joined in a cube is already aggregated, it prevents analysts from getting a complete picture of all their data. This post will cover the limitations of traditional cubes and how Analytical Views provide more agile methods for analytics while preserving the performance benefits of cubes.
Another drawback to OLAP is that there are several time-consuming steps to a traditional BI-on-cube approach. First, you have to identify the dimensions that define a business problem, and then build the cube based on this information. For example, an OLAP cube for reporting sales might include Salesperson, Sales Amount, Region, Product, Month, and Year dimensions. Then, using the cube table as a data source, you can finally start to build the BI reports.
This approach has several drawbacks:
Laborious definition process. You must first identify all dimensions before building out the BI application.
Difficult to change application requirements. Since you can’t access other dimensions within the cube, you would have to rebuild the code if you want to change application requirements.
Limited drill-down capability. Cubes by nature have limited drill-down capability since data is summarized. If the source data table(s) are on a different system, drill-down to the raw data is difficult if not impossible.
Arcadia Enterprise is a native visual analytics platform that is completely native to Apache Hadoop (and other big data platforms) and resides “in cluster.” (Note that Arcadia Instant, the free downloadable product, does not provide the enterprise-scale features described in this blog.) Here’s how the Arcadia Enterprise architecture helps with the issues listed above:
Work with a single, unified view of the data. The Arcadia Data Analytical View capability enables a much faster provisioning of queries. Analytical View flips the cube-based BI flow on its head by tracking ad-hoc query behavior across the end users who design and consume the visualizations in Arcadia Enterprise. The beauty is that end users don’t need to create and continuously refine their own cube. Analytical views kick in automatically, and can optimize for joins, distinct counts, medians, etc. Because the BI applications are built against the base data, the application developer works with a single unified view of the data with access to all fields even though specific reports may be supported by different analytical views.
Eliminates the need to build datamarts/cubes or extracts before analyzing the data and developing BI applications —start exploring raw data immediately. No need to submit a ticket to IT to build a cube first.
Get full drill-down to the finest grained data contained in the base table, allowing better answers to more detailed questions which can be incorporated into the BI applications. The Analytical View capability enables you to connect right to your data to continuously explore, model, and refine it. You can go straight into big data using our intuitive drag-and-drop self-service interface, which provides exploration and semantic modeling on the breadth and depth of all your data.
Easy to access and scale. Arcadia Data unifies visual analytics and BI on Hadoop, so it’s easy to access big data in Hadoop, Apache Solr, and other data platforms. The Arcadia Data visualization platform uses two levels of acceleration technology: 1) Web server local cache of database engine results from front-end requests, and 2) Analytical View support within ArcEngine to service query requests from pre-aggregated results rather than base tables.
Our on-cluster execution engine delivers scale-out performance, with an active data layer for self-learning optimization and advanced analytics and an intuitive drag-and-drop self-service access interface, giving you the ability to create a full range of charts and graphs.
Lower TCO. Frees up IT resources that legacy BI tools require while lowering infrastructure costs associated with traditional BI implementations. Since Arcadia Data is a completely web-based application, it doesn’t need a local client or data deposition tools.
There are many ways Arcadia Enterprise can be used to provide organizations with the powerful and flexible visual capabilities of a seamlessly integrated native BI Hadoop platform. Here are just a few:
MarketShare Partners uses Arcadia Data to provide large-scale digital marketers better decision analytics for online and mobile campaigns. Users can rapidly drill down directly to raw records for fast, Hadoop-native visual analytics. They were able to switch from one-off static reports to a multi-tenant environment that delivers fresh marketing attribution data and embeddable customized reports in hours versus days.
Kaiser Permanente is deploying Arcadia Data across the entire organization, giving business users the ability to perform exploratory analysis using a drag and drop, browser-based interface and to create production-grade data-driven applications with sub-second performance.
A leading data infrastructure company uses Arcadia Data to securely share data with appropriately authorized users via Hadoop’s native security services, gain continuous interactive data granularity, and rapidly propagate discoveries.
Arcadia Data provides streamlined, easy-to-use visual analytics, BI and data discovery, all in a converged platform that directly taps into the power of native on-cluster Hadoop processing. By delivering secure, accelerated access without first needing to build data cubes, Arcadia makes it easy for you to create interactive, self-service data-driven applications and iterate insights faster.