Does this scenario sound familiar in your organization? You are collecting so much data and information from “events” like customer wait times, machine failures, information reports from social media, IoT information like GPS recordings or temperature monitors and it’s all just sitting around collecting dust. You have so many data sources generating information in so many different languages, you just are unable to turn this data into action.
Traditional data processing doesn’t help much in this “do-it-now-world.” Because collecting information, storing it in a relational database or Hadoop cluster, and running queries on it to discover meaningful events is just too slow. Your opportunity to influence the right outcome on all your data quickly slips away. And, the events you want to track are happening frequently and close together in time and the event needs to be detected and responded to quickly.
That’s why your organization needs event processing. It’s often referred to by many names: real-time analytics, streaming analytics, complex-event processing (CEP), real-time streaming analytics, and Kafka Streams. If we break down the term “event processing”, an event is a data point triggered by a change in state or an action taken place within the business (an event). For example, a bank transaction is an event. An “event stream” is a sequence of events ordered by time.
Analyze data immediately after it is created
Today, businesses need to be event driven, but the reality is, most are not yet. The magic of event stream processing is that it connects to all of your data sources, normalizes, enriches and filters the data and automatically applies rules to the data to reveal patterns, relationships or trends in real time. This means you can quickly analyze data at an almost instantaneous rate after it’s created.
You can then begin to correlate events from multiple data streams and data repositories and over time, you can see patterns emerge that describe events you care about. You can add contextual data from various sources to ensure proper interpretation of events and then you apply real-time business logic and rules or machine learning to trigger an action. Action could mean invoking an app or starting a process, storing data in a persistent ledger or streaming data to a dashboard.
Get contextual offers to customers
For instance, if you are a retailer, with event streaming architecture, you could detect where a customer is near a retail outlet, understand that customer’s previous engagements like web page views history, abandoned shopping carts, or social media activity. Armed with that information, you are able to make a contextual offer to entice the customer to come into the store and make a purchase at the exact time they are near the retail outlet.
Event processing lets you predict and act while the value of your data is still high. It allows you to turn your data into action—and empower business users to define the rules that take the action that gives you a competitive advantage.
Identify next best actions
And, with machine learning as an option you can not only identify the next best action but continually learn and improve business rules. The key to successful event stream processing is processing events in real time to identify the next best action. Learning and improving in a continuous loop.
Automate business processes
Event stream processing can make sense of vast amounts of data arriving at great velocity into your business to help filter out what’s important so you can automate processes. In fact, in a recent HBR survey, building more intelligence and automation into business processes was cited by 80% of respondents as very important to the success of their enterprise.
As you can see in the diagram below, the example we used is with data streaming into a bank to show how event stream processing can help automate processes. It works like this:
- Data streams in at high velocity and high volume (millions of events)
- Data goes through event stream processing (ESP)
- AI/ML are used to decipher patterns, trends, and anomalies
- From there, depending on what the event is, the system can be set up to automatically run certain actions
For instance, if it’s a small fraudulent event, you might just invoke a BPM process to alert the user. If it’s a larger one, you might set up the system to send a text automatically to the user. If it’s a really large event, you might invoke the CRM to put a freeze on the account.
Event stream processing allows users to automate actions based on events happening within the organization. This could be used to improve operations, advise on how best to engage with customers at any given moment or even to take steps to mitigate potential risk. Basically, event stream processing coupled with AI/ML, is a way to parse through all of your data fast enough to detect patterns and automate actions.
Apache Kafka for streaming data
Currently, Apache Kafka is the most popular tool for streaming data. We can use that streaming data to turn that data into action. That’s why we designed TIBCO event processing to work with Apache Kafka in a simple to use, point and click deployment.
TIBCO event streaming architecture enables responsiveness to match the velocity of your business. And it’s a smart and aware solution that learns and adjusts to circumstances. TIBCO provides the complete end-to-end solution. From commercial support for Apache Kafka, to open source tools like Project Flogo that provide integration flows, stream processing and a business rules engine to embed machine learning models to quickly turn streaming data into action.