AI Insights
Showing results for 
Search instead for 
Did you mean: 

Up the value of your analytics with a better platform architecture


An emerging high performance computing model is helping enterprises break down the wall between operations and analytics. Take a look at the potential of a better platform architecture here.

Blog_htap-platform-architecture-1.jpgImagine a computer programmed to sell a certain stock when the price hits $80 a share. Now, imagine the information received by the computer is on a 48-hour delay.

In a world where financial transactions are processed in microseconds, this scenario sounds ridiculous. Make no mistake, no brokerage could operate successfully under such conditions. But this scenario—a lengthy delay between business intelligence and business action—is common in many industries, like retail, where a wall often exists between analytical and transactional systems.

An emerging class of in-memory computing solutions known as hybrid transactional/analytical processing (HTAP) aims to break this wall, allowing organizations to respond to business intelligence in real time.

Defining a platform architecture for HTAP

Gartner Inc. defines HTAP as an application architecture through which concurrent analytical and transaction processing algorithms share the same data. The architecture can have a strategic and transformational impact, redefining the way business processes are designed and executed.

Gartner divides HTAP into two different styles:

  1. Point-of-decision HTAP: In this model, transaction processing and analytics are segregated into two separate applications, allowing advanced analytics to be performed on "live" transactional data—which is difficult to achieve in traditional analytics setups. This style of HTAP is best for building knowledge and delivering tactical quick wins.
  2. In-process HTAP: This model weaves together real-time analytics and transaction processing techniques in the same application, so analytics can directly influence the execution of business processes. This style of HTAP can deliver greater transformational benefits to an organization.

Traditional data management technologies can power some HTAP applications, but generally, HTAP solutions require in-memory computing technologies. With in-memory computing, operations all happen in-memory with the system continually writing data to disk. This eliminates the I/O and ACID transaction requirements of online transaction processing (OLTP) applications and speeds up data access dramatically. Unlike data stored on disks, RAM-stored data is available instantly.

HTAP solutions bring benefits to a variety of industries

Gartner identifies the following potential and real-world business uses of HTAP:

  • Supply chain planners can take advantage of the "in-the-moment" availability of key performance indicators (KPIs) to quickly address supply disruptions or fluctuation in demand. For instance, if data showed a company was on track to sell less of a product than previously predicted, the planners could cancel their orders early—preventing excessive inventory or penalties from cancellations.
  • Real-time data can help fulfillment managers decide how to prioritize the scheduling of purchase orders. In-process HTAP applications could even use analytics to prioritize and execute these transactions without human input.
  • A financial institution could use HTAP solutions to introduce innovations, including customer self-service analytics and real-time risk management capability. This can help the bank differentiate from its peers and establish a competitive advantage in the marketplace.
  • Using HTAP solutions, an online gaming company could respond to player-generated data almost instantaneously, helping create a customized gaming experience and improve outcomes—including retention, engagement, and monetization.

Leverage HTAP for more advanced analytics

While the potential rewards of HTAP solutions are great, the implementation challenges are also considerable—including those around architecture, technology, skills, and cost. While in-process HTAP has more transformational potential than the point-of-decision method, it's also more difficult to implement. This is why Gartner predicts that at least 75 percent of HTAP projects will adopt a point-of-decision approach in 2018.

Because HTAP is still in its infancy, Gartner warns that industry expertise is limited to IT professionals in leading-edge organizations, and best practices for the architecture haven't been established yet. Retrofitting applications for in-process HTAP requires extensive—and potentially risky—reengineering efforts.

Organizations need to familiarize themselves with HTAP and in-memory compute through conversations with technology service providers, followed by the implementation of some low-hanging-fruit point-of-decision applications. After demonstrating the value of these initial projects, IT leaders can build a business case for more advanced forms of analytics.

When the walls between processes and analytics in the enterprise are broken down, the potential for business transformation can reach full capacity in your enterprise. Register here to read more about enabling digital business innovation through HTAP platform architecture.

Featured article:  The Machine seeks adventurous developers


0 Kudos
About the Author


Calvin is a freelance business and technology writer. His work has appeared in the business section of The Boston Globe, BizTech magazine, Yahoo, and elsewhere.