Enabling Your Advanced Analytics Environment

Intel Corporation outlines how advanced analytics can fit into today’s – and tomorrow’s – IT environments, and explains why now’s the time to experiment.

 

You probably already know your business is sitting on a gold mine. As the Internet of Things (IoT), social media and smart devices play an increasingly prominent role in most business environments, the data they create is growing at an unprecedented rate (IDC predicts that we’ll have created 180 Zettabytes of data by 2026, up from just 10 ZB in 2015). But while converting this rich stream of information into business-driving insights is a big challenge, with emerging technologies like artificial intelligence (AI) and 5G networks, it’s within IT’s power to make it happen.

If you’re not already thinking about it, now’s the time to begin exploring how you can move beyond traditional business intelligence (BI) to real-time analytics in order to boost efficiency, improve security and drive innovation – all of which ultimately will help make your business run better and your customers happier. A number of organizations are already doing this and reaping the benefits.

Where to start?

In my conversations with customers across industries, I’ve found that most IT shops (and many business units too) have a high level of interest in how advanced analytics, AI and machine learning can positively impact their business. There’s a lot of enthusiasm out there! However, taking the first steps can be daunting. How do you turn that enthusiasm into a tangible analytics-ready IT environment?

The first step is to make sure you have a strong data strategy in place. What data do you have access to? Who needs to use it? How are you going to capture, keep, store and analyze it? Traditionally, this has been done using data warehousing, which works well for manageable, relatively structured data sets. When you need to handle data on a vast scale, though – like that generated through your company’s social media accounts, for example – an enterprise data warehouse quickly becomes cost-prohibitive.

By using cloud and open source data storage, and processing platforms like Hadoop, it’s now possible (and relatively affordable) to handle these larger data volumes.

Choose the analytics right for you

Once you have your data strategy in place, you can explore the analytics possibilities. Speak to your business units about what questions they’d like answered to ensure the analytics environment you’re building aligns closely with the company’s strategic goals. There are five main types of analytics to consider:

  • Descriptive Analytics is the simplest type of analytics. It tells you what has happened, and it’s what BI tools typically offer
  • Diagnostic Analytics tells you what happened and why
  • Predictive Analytics looks ahead, assessing what will happen, when and why
  • Prescriptive Analytics enables simulation-driven analysis and decision making
  • Cognitive Analytics, the most complex, enables computerized thought simulation and actions. This is where AI comes in, using machine learning to automate decision making and actions

Each type of analytics can add value to your business, but the more complex you get, the more specialized your infrastructure needs to be. For example, cognitive analytics tends to require a machine learning platform such as the Saffron Technology™ platform from Intel, as well as silicon specialized to support large workloads, vast scalability, and multiple data sources.

Analytics Solution Stack
The Analytics Solutions Stack (Click image to enlarge)

Investing in a whole new infrastructure isn’t realistic for most of us though, but this isn’t a problem when it comes to analytics. As with many IT projects, it’s a good idea to start small and build up. Use your existing infrastructure to experiment with the less complex analytics models and demonstrate their value to the business. Once this is done, it’s easier to secure further investment for scaling up and advancing to prescriptive and even cognitive analytics over time – if appropriate.

Layer up

When building an analytics infrastructure it’s best to think about the solution stack as four complementary layers, each including a variety of technologies, depending on your own organizational needs, legacy systems, and preferences, and the type of analytics you’re going for.

The Infrastructure Layer: This is the foundation that will enable you to acquire, store and protect your data, and to run commercial and open-source analytics. A typical infrastructure layer may support open source distributed processing frameworks, non-relational analytics databases, and analytics applications.

The Data Layer: Pre-analytics, this layer may have been comprised mostly of relational databases. With more streaming and unstructured data now, this layer may be supplemented by the Hadoop Distributed File System, holding your enterprise data hubs or data lakes. The data may also sit in NoSQL databases, your existing enterprise resource planning (ERP) system, or streaming data from IoT devices that feed real-time analytics environments like SAP HANA.

The Analytics Layer: Data scientists use tools such as HP Haven* Predictive Analytics and PredictionIO* here to convert data into information that can be used by the company’s end-user analytics applications.

The Applications Layer: Open source or commercial software, such as those provided by SAS, Tableau and Qlikview, provides tailored analytics capabilities to different types of user and industry.

These layers enable you to take advantage of your existing data management systems as well as newly implemented technologies. Additional technologies like AI and performance or security solutions can be applied across the whole stack to accelerate insights and strengthen data protection.

Whether you’re just starting out on your analytics journey or preparing to build on what you already have, I wish you success. It’s a challenge, but a highly rewarding one. To help you on your way, check out Intel’s ‘Getting Started with Analytics’ Planning Guide, which goes into more detail on the types of analytics out there, the solution stack, and provides some handy practical tips.