Components of a Fast Data Architecture

How does an organization differentiate itself in the emerging digital world where data older than an hour is no longer helpful in making decisions that will deliver customer-pleasing results based on appropriate actions? Well, you could build a Fast Data architecture yourself, you could wait for an alliance to agree and deliver, or you could buy one out of the box. Some organizations want  a designed-for-purpose and integrated approach that is proven to work today by combining the integrated source of signals and patterns, sense and alert facilities, technology-assisted human insight suite, and automated action capabilities, all based on faster big data infrastructure. Let’s dig into each of these contributing stacks.

Integration of Data Sources:

Managing the multitude of signal emitters, whether they be heterogeneous hardware devices, or heterogeneous software activity is crucial at the mouth of the data funnel that selects, condenses, analyzes, and takes guided or independent actions based on organizational goals. This requires a rich integration and aggregation stack that can handle the variety and complexity of signal sources that generate data and event data. 

Real-Time Intelligent Technologies:

When the speed of the data increases, new capabilities need to be there to deal with the speed increase and get the most out of the data streams. In order to sense patterns of organizational interest, opportunities or threats, there needs to be a complex events capability that can recognize predefined patterns. In the case of an unanticipated event or pattern, real-time analytics can translate an ordinary event into a pattern to watch and add to the list of threats or opportunities. In the case of anticipated or unanticipated, alerting mechanisms need to be available to gain attention. In the case of anticipated patterns, there can be inventoried actions.

Human Insight Assistance:

In this class of capabilities lie a significant set of features to support prediction and decisions. This requires assists for humans to visualize conditions, predict the next best action, and create action or task lists (with automated responses where anticipated), or for humans to deal with the unknown. The key here is the power of the visualization and the ease of use. Some single events or patterns do not need a response, but decision activity may be involved in either case. 

Guided and Automated Actions:

Decisions and actions are not only important in the context of a pattern but are also important in the context of an overall end-to-end process or a simple set of tasks, or both. The context of channel and customer needs and desires—that are dynamic by nature—need to be factored into appropriate actions. This requires support for a variety of process styles from straight through processes to dynamically changing and contextual processes. It may even require kicking off a longer running case that needs deep knowledge requiring great collaboration.

Net; Net:

As you can see, Fast Data architectures have to include more than support for data. They have to support visual decisions and smart actions. All of this has to work seamlessly and flawlessly, taking out as much complexity as possible. There are a number of vendors who have or are pursuing speedier big data architectures and are at various stages of maturity. Many view fast data as a better way to stay competitive. 

See http://jimsinur.blogspot.com/2015/08/fast-data-is-better-than-big-data.html

A Sample Integrated Architecture from Tibco that is available and expanding today:

http://www.tibco.com/blog/2015/08/26/tibcos-fast-data-architecture-is-built-for-speedy-results/