Why an Event-Native Mindset is Now Essential for Data Architecture
In Gartner’s Symposium Presentation “Strategic Trends in Application Platforms and Architecture” Distinguished VP Analyst Yefim Natis proposes that it is not enough for businesses today to be “ready to change,” but instead, companies need to be “ready to act” in real-time and understand the context of the action. Simply being prepared to respond to change is not sufficient in today’s fast-paced and constantly evolving business environment. Organizations need to be able to anticipate and proactively respond to changes in order to stay competitive.
But how can a business create a data architecture that supports ‘ready to act’ applications and systems? Gartner suggests that by embracing an Event–Native Mindset and an Event-Driven Architecture organizations can create an enterprise nervous system that delivers continuous intelligence and keeps the business always ready.
Applying an Event–Native Mindset to Data Modeling
Consider for a moment that everything that happens within a business environment is an event, and every event impacts an entity or is carried out by an entity. An entity in this context is a core business concept like a customer, patient or product. Everything a patient does – getting diagnosed, visiting a specialist, being discharged or receiving a prescription – is an event. Anything that happens to a business’ customer, from making purchases, returns or calling for support, is an event. When products are created, tested, and updated, these activities are also events.
Recommended: John Snow Labs Puts Responsible AI to Practice with the Release of the…
By adopting an Event–Native Mindset in data modeling, organizations can develop more agile and responsive data architectures that are better suited to handle complex and rapidly changing data environments. This approach can help organizations to more quickly identify and respond to changes in data patterns, and to more effectively leverage the insights and value that can be derived from event-driven data models.
Centering data modeling around events greatly simplifies the data modeling demands of almost any use case, from deploying knowledge graphs to data fabrics. Simply assign important entities (like callers for contact centers, patients for hospitals, and customers in financial services) to data models, then whatever happens to that entity is modeled as an event with a timestamp, description, and specification of the outcome. This simplified schema is applicable to data of any source and primarily consists of these two objects.
The benefits of this model are considerable. Its simple data shape makes it easy to query—which reduces costs, time, and human effort for this task. Subsequently, users can rapidly retrieve 360 degree views of entities and extract entities as a series of events. The latter is useful for event prediction with machine learning, so organizations can understand next-best actions, recommendations, and prepare for the future. Event-based data models also minimize the complexities of processing real-time data, which is why many streaming data platforms and services are predicated on an event-driven architecture.
Top AI ML Insights: 5 Biggest AI Tools You Should Be Using Right Now with ChatGPT
Simplified Data Shape
The querying advantages of an entity event model deliver immediate value to organizations. Since the schema consists of two objects, the data shape is simple and the same for data of any variety, domain, or application. The simplicity of the data shape makes it easy for business users to understand the data. With alternative state-based modeling approaches the schema for a data warehouse would take up entire walls if visualized with something like Unified Modeling Language (UML). People could spend weeks deciphering what’s in those tables. Conversely, the information in entity event models is comprehensible to even laymen in just an hour’s time.
With this simplified schema, queries become much more straightforward, less time-consuming, and less resource-intensive. For example, a SQL query to find patients hospitalized due to cardiac ailments from 2010 on would take up to three pages. With the entity event model, the same query would be a few lines, which reduces its time-to-value—and costs.
360 Views
Another distinct benefit of entity event models is how easy it is to obtain 360 views of entities and perform analytics on the results for clinical care research, personalized marketing or sales opportunities. Entity identifiers that uniquely identify an entity are stored in every event, which enables a user to retrieve everything about the entity in just one API call.
Simplifying Feature Extraction for Machine Learning
The ease of querying entity event models is optimal for data science use cases, specifically the feature extraction for machine learning applications. For example, users can extract a customer’s purchase history as a series of events to determine which products or services he or she is likely to avail themselves of next.
Organizations can capitalize on this information with personalized offers for both up-selling and cross-selling opportunities. Furthermore, analyzing entities as a series of events naturally lends itself to time series analysis, which further enhances event prediction. Medical practitioners, for example, can identify patients’ susceptibilities so they can prepare for—and mitigate—them.
The Future
In terms of ease of use, querying simplicity, and advanced analytics implications, few data models can match the utility of an entity event model. Conceptualizing, modeling, and acting on data as temporal events is the foundation of this approach, which is being readily embraced by an increasing number of facets of data management.
Comments are closed.