The ubiquitous phrase "big data" has worn out its welcome. For most businesses, the emphasis lies in useful analytic results, not the size or the source of data that produces insight. But that doesn't mean analytics technology should stand still.
On the contrary, argues Christian Gheorghe, CEO of Tidemark. He maintains that legacy EPM (enterprise performance management) applications restrict the benefits of analytics to a chosen few in the organization. Tidemark's cloud-based analytics platform seeks to open up this exclusive realm and create what Gheorghe calls a "companywide culture of performance." In this week's New Tech Forum, Gheorghe contends that his platform delivers just that, through a combination of object-oriented application models, cloud architecture, and self-service user experience. -- Paul Venezia
How cloud-first EPM yields enterprise-wide insights
Fueled by the explosion in data and the consumerization of IT, today's analytics apps are simple and intuitive, and they're designed to provide much greater insight into both business performance and new revenue opportunities. A key goal of these apps is to enable the entire organization to positively impact business performance. By engaging more users in forecasting and planning, for example, the data is more contextual and readily accessible to multiple lines of business.
Conversely, database-centric legacy architectures of EPM applications prevent businesses from easily exploiting varied data sources, while their design restricts usage to power users. As a result, only 12 percent of employees use their company's analytics tools today, creating a handicap that many businesses fail to identify.
The ETL (extract, transfer, load) model developed back in the '90s remains a commonly used method to extract predetermined subsets of structured data from the monolithic transactional systems that serve as data repositories. Once extracted, the data is transformed into a structure that makes business sense and loaded into business apps.
The problem is that data today comes from many sources and not all of it is structured. Nonstructured data (documents, RSS, social media, sensor data) is now essential to the analytic process. In the legacy world, the most common way to make it available to users is for IT to work in conjunction with finance to force big data into a cube or transactional system -- an expensive and time-consuming process that requires compromises.
Modern EPM solutions turn the traditional concept of ETL on its head, prioritizing users and the business questions they want to explore. They forge entirely new models as needed to extract the data and answer each question in real time. Making the process collaborative gives users the ability to analyze data within the specific context of their business problem. Instead of ETL, the more optimal approach is ELT: First, extract all the data from sources; then load it into the computational grid in the cloud; finally, transform it as business users are asking questions and modeling processes.
Object-oriented application models
For optimal results, everything a user can configure in the model should translate to an object or attributes of the object. These objects have complex, many-to-many relationships with one another that define how the application will transform the data. For example, a company's model may contain 10 different dimensions, including product, customer, region, and time. As the user changes the configuration of that model -- say, by adding a new hierarchy or a new dimension -- the configuration is translated into transformations that will be applied at runtime.
With this approach, a large array of use cases can be quickly configured, tried within various scenarios, and modified as the business changes, all without the normal hand-holding required from IT.