What Is Real-Time Analytics?
As business decision-makers search for deeper and timelier insights, real-time analytics sometimes called operational intelligence is gaining traction across all industries. These tools gather actionable intelligence from both historical and real-time sources that stream into an organization continuously.
Real-time analytics is gaining fame for its ability to turn data into insights within seconds or sub-seconds. But a simpler way to think about real-time analytics is having information immediately available after it’s collected for analysis, reporting, and decision-making by whoever needs it—often proactively. Your operations team might not have realized that a device was about to fail but predictive analytics will surface this time-sensitive information.
How is this achieved? First consider traditional structured databases such as SQL/BI systems. In this approach, information is stored, indexed, and subsequently processed by queries. This was adequate in days before Twitter posts could launch a fashion trend or retailers could lose a customer if they didn’t push timely or appropriate product suggestions to customers at peak buying times.
The relentless pace of business is causing companies to take a new approach on data velocity, the speed at which data is processed. Rather than storing and indexing information in traditional databases, the use cases of real-time data require data to be processed “in-flight” as it streams over a server. The real-time data must then be integrated with historical data.
Most companies collect an enormous amount and variety of data, but they’re unsure how to utilize it or combine it with internal or external data to gain timely insights about their brand, the customer experience or market trends. Real-time analytics provides a way to unleash the power of all that data, and allow business decisions to be made at Internet speed.
What Is the Business Case for Real-Time Analytics?
Retailers, manufacturers, financial services firms and a host of other industries are struggling to keep up with the pace of data that is generated and must be processed and analyzed to keep ahead of customer’s ever-increasing demands and competitive pressures.
Typically, the value of data decays significantly over time. By operating in real-time, the data quality is not adversely affected, and analytics can be applied to business processes, which have narrow time windows or where changing events require immediate response.
As the influx of data from sensors and the Internet of Things (IoT) increases, this process has become more imperative, since the value of this data can evaporate within days, hours, minutes or even seconds. For example, IoT data that directs a driverless truck is the perfect example of information that becomes worthless – or even dangerous – with the slightest latency. Likewise, data that indicates fatigue in a machine on a manufacturing line becomes useless once the machine fails.
Real-time analytics addresses many organizational pain points. Online retailers are blending transactional and web browsing activity to determine the next best offer to serve up to a customer. Banks are analyzing behaviors to determine fraudulent activity or detect signs that a customer who works with one of their departments is ready for a pitch from another department. Dynamic pricing, risk management, call center optimization and security are just a few of the processes that can be optimized with real-time analytics.
In these cases, real-time data allows companies to deliver value-added services and products at the very moment the customer wants them, and to defend against negative consequences before they become devastating. Real-time analytics can separate trends from noise. As the pace of business grows ever more relentless, the early warning systems that real-time analytics propel will at first become key differentiators, and then soon necessities that provide the kind of service that customers will expect.
How Do Real-Time Analytics Fit in with an Overall Analytics Strategy?
Analytics is a spectrum, with most companies adopting a mix of analytic approaches based on data types, workloads, and the type of business problems they are trying to solve. Analytics now spans five categories: descriptive, diagnostic, predictive, prescriptive, and cognitive.
Descriptive analytics answers questions about what happened in the past. Diagnostic analytics offers insights about why those events happened. Predictive analytics analyzes current and historical data to provide insight on what might happen in the future. Prescriptive analytics suggest actions an organization could take based on those predictions, while cognitive analytics automates or augments human decisions.
These five categories build on each other in a stepwise manner, moving an organization toward an on-demand enterprise where decisions become faster and better.
Predictive analytics are the beginning point of “advanced analytics,” where decision-making may be fueled by real-time information. Predictive analytics, then, is a use case that benefits from the capability of real-time analytics.
No matter the type of analytics companies employ, they need to adopt a comprehensive data strategy built upon a modern infrastructure that breaks down both data silos and organizational ones too. The common theme is the ability to capture, store, analyze and secure data so that insights can scale rapidly across the organization to allow timely business decisions.
What Infrastructure and Skills Needs Are Raised by Real-Time Analytics?
The analytics solution stack is made up of four layers—infrastructure, data, analytics, and application. Intel® technologies span every important part of a company’s infrastructure, across the network, storage and compute, allowing data to be efficiently managed and rapidly harnessed for competitive advantage. A consistent architecture for example one based on Intel® Xeon® Scalable processors across an organization provides a predictable path to rapidly scale analytics initiatives without the need to support multiple architectures.
Traditional big data solutions, focused on data warehousing, are not suited for most real-time data processing. Increasingly, cloud vendors are providing Platform-as-a-Service (PaaS) and Software-as-a-Service (SaaS) offerings that can be leveraged in the service of real-time analytics. Brokered solutions across clouds enable companies to run workloads wherever they wish depending on the volume, variety and velocity of the information.
As companies generate huge amounts of data in the cloud, they must determine which data needs to be moved back to the enterprise to make intelligent decisions. Real-time data may be processed on “the edge,” with data analysis occurring at or close to the collection point. However, real-time analytics in the data center requires rapid access to and analysis of increasingly large amounts of data. This means it is essential to optimize every level of your infrastructure from the CPU to memory/storage subsystems. Persistent memory technologies keep more data closer to the CPU and retained in-memory during power outage cycles, eliminating the latencies caused by I/O bottlenecks, fetching data from slower SSDs and speeding up restarts.
Real-time analytics requires taking data from anywhere, in any format, and getting it into the right record form so that it can be processed as a whole. The key is to understand where data is created, and how it will be used to improve business processes and decision-making.
Who Are the Key Players in Real-Time Analytics?
As an analytics technology partner, Intel provides the flexibility to choose from industry-leading analytics software solutions that are either open source or proprietary.
SAP HANA* is a single database that combines a database with advanced data processing, application services, and flexible data integration services. HANA leverages in-memory database software, an approach to querying data when it resides in the system’s memory (today called RAM) rather than querying data that is stored on physical disks.
This allows customers to process data in many new ways much faster and build a series of what-if scenarios to help exploit opportunities or avoid problems. Other traditional technology vendors, such as IBM and Oracle, have also enabled real-time operations in their platform with new technology.
Open source solutions, centered on Apache Spark* base code, bring real-time analytics to unstructured data such as social media, images and video. Spark uses in-memory analytics scaled across numerous systems so that large amounts of data can be processed in parallel.
Many of these solutions can be offered in the cloud, allowing analytics to be run where such data as social media and the Internet of Things is being generated. As a result, companies can query transactional and online data to shed light on patterns and trends in real-time, moving as quickly as the world and their customers do.
The number and variety of solutions, and new market players, are entering the market constantly. This provides a rich ecosystem of solutions primed to take advantage of the compute, network, and storage capabilities that Intel provides to enable ever-increasing nimbleness in enterprise analytics and decision-making.