Real-time analytics provide a strategic view of your business, second by second. Hybrid transactional/analytical processing makes these insights possible for your enterprise with in-memory computing.
Close your eyes and ask yourself how many times in the past decade you've heard "real-time" terminology in your enterprise. Now, open your eyes and look around. Have the real-time insights and analytics you've heard about met their potential yet?
Despite advancements in in-memory computing, businesses are still learning how to analyze transactions on the fly. How can enterprises fully unleash the potential benefits of real-time analytics? They can start by marrying transaction processing and analytics.
Close the transactions and analytics gap
Traditionally, analytics and transaction processing are sequenced activities targeting different use cases. Transactions are processed in real time as they hit the database; conversely, analytics take historical data and find patterns in it to provide actionable intelligence to business executives. Working on data across a historical window can provide useful information about trends, but it won't help companies make immediate decisions based on up-to-the-minute conditions.
Because of the nature of these processes, the wall between transaction processing and analytics has proven hard to break. Companies typically use a primary transactional data store to handle operational transactions and a separate data warehouse for analysis and business intelligence.
What if the barrier between these two types of processing could be collapsed? Analytics—using the same data as transactional systems—could provide more current insights and enable a business to react faster to a changing market environment. That's where hybrid transactional/analytical processing (HTAP) steps in: It allows companies to use the same data for real-time analytics as they use for transaction processing—or, at least, an immediately generated copy of that data.
Choose between two HTAP deployment models
HTAP removes the wall between online transaction processing and analytics, but it brings new challenges, such as in-memory data processing. To support in-memory computing, you can turn to servers that are purpose-built to handle the most demanding and critical business processing needs, giving computers the absolute fastest performance they can deliver.
That said, the business scenarios for HTAP depend on how a company deploys it. There are two kinds of HTAP deployment, each offering a unique business benefit:
1. Point-of-decision HTAP deployment
In a point-of-decision system—the easier of the two deployments—both transaction processing and analytics rely on a common in-memory data store, but the two processing applications don't talk to each other. The transaction processing system, such as a sales processing or fleet scheduling application, will rapidly iterate through transactions in memory. The analytics application will also draw on that data to provide real-time insights.
Consider the real-world impacts those insights could provide for an enterprise. For instance, the analytics program might search for trends in online sales it could use to advise inventory managers to order more of a product quickly. It may even be able to recommend changes to product pricing based on real-time patterns it sees in transactions. This could directly translate to an increase in sales.
2. In-process HTAP deployment
While a point-of-decision HTAP system gives business managers better visibility into business conditions than ever before, an in-process HTAP setup takes things further. Instead of operating separately inside their own silos, the transaction processing and analytics systems talk to each other directly and make decisions about transactions based on real-time intelligence.
This can lead to huge business wins. For example, what if an analytics system noticed a spike in service calls in a geographic region, which adversely affected the scheduling of maintenance visits? With the ability to influence transactions, the system could suggest new routes and schedules for field engineers. An analytics system that noticed an emerging pattern of fraudulent transactions could steer the online transaction system to quarantine certain orders until staff could examine them closely.
Start your real-time analytics journey
Shifting to an in-process HTAP environment will involve some major reengineering of existing business processes and transactional applications to facilitate a real-time analytics system and make decisions based on its recommendations. In some cases, it could entail a complete redesign, so you'll need to carefully consider the costs and risks involved and navigate thoughtfully.
Point-of-decision HTAP is an easier proposition—one where transaction processing applications don't need to change, because they're not talking directly to real-time analytics systems. Deploying point-of-decision HTAP first gives companies the chance to realize ROI by identifying a few quick wins from real-time analytics—and this might be the perfect place to start your journey. Then, your team could begin building out the necessary skills to tackle a more complex in-process deployment with confidence.
The concept of real-time data analysis has definitely been around for a while, but it's becoming more viable for companies as in-memory computing advances. Now, your enterprise needs the business vision and technical skills to act on the opportunity. To read more about enabling digital business innovation with real-time insights, register to read Gartner's HTAP paper.