4 things to consider when setting your fast data strategy
A new Forrester study, Part 2 of 2
In the last post in this series, we highlighted a new Forrester Consulting study commissioned by IBM in October 2018 on fast data, the omnipresent, technically challenging class of data that many enterprises today are looking to use.
Many of the respondents reported problems getting the most from their data, primarily due to insufficient and poorly integrated tools. Fortunately, there are strategies and technologies that have the potential to solve the challenges of fast data.
Fail to plan, plan to fail
In the study, the definition of fast data starts with the technical characteristics mentioned in our last article, but there’s more to that definition. Fast data is generated, streamed, stored and immediately analyzed “by applications that detect patterns, automate decisions, and immediately initiate actions”. So while fast data is technically challenging for all the reasons we discussed in part one, that challenge can be met with automated, intelligent applications.
A smart strategy is the key to setting up an analytics environment that gets maximum value from fast data. That’s why almost 90 percent of Forrester’s respondents say they will adopt or reevaluate a fast data analytics strategy within the next two years.
But not all strategies are created equal. What’s most important for fast data success? The Forrester study digs into the many responses and comes up with some key considerations.
1. Storage and compute requirements are different for fast data than for traditional analytics and require different expertise.
The storage and compute infrastructure used for report- and dashboard-driven analytics probably isn’t suited to fast data. Choose scalable infrastructure and “consult with data science, AI engineering, and application development teams to forecast storage and compute needs for next-gen, real-time digital services.” The infrastructure must be part of the strategy, not just something chosen by IT.
2. Data and application architectures should support real-time services.
“Fast data” insight means more than just capturing and analyzing data in real time. The data should also help drive real-time services. As the study puts it, enterprise architects must “evolve existing data and application architectures to enable application developers, data scientists and AI engineers to create intelligent, real-time services”. This large-scale goal should be part of any successful fast data strategy and may be why 73 percent reported using machine learning for streamed and stored data.
3. The priority for fast data solutions must be on integration and simplicity.
Two truths that stand out from the study results are that fast data can sometimes create unpredictable demands, and some fast data solutions are poorly integrated. This suggests that better integration could have enormous benefits. A data strategy should prioritize ease of use, robustness and interoperability.
For example, a hybrid architecture that includes both on-premises and cloud components to deal with both structured and unstructured data is great unless you can’t get the various components to play well together. If you’re not building your own platform (and this report suggests you shouldn’t), look at data platform providers that can offer a range of solutions designed from the ground up to work together.
4. All these strategic goals can be aided by the right technology.
The right technology is essential to fast data success. For example, well integrated machine learning capabilities help enormously to provide real-time, scalable, event-driven steaming applications. An open source foundation for a data platform along with an open data format can have many advantages, from simpler integration with other ecosystem components to avoidance of vendor lock-in.
It even can help reduce personnel demands, because administrators and developers may already have the relevant open-source skillsets rather than needing to learn a whole proprietary system. Make sure your strategy accounts for the latest technological resources to optimize fast data and analytics systems.
Acting out best practices
The Forrester study shows that some enterprises are already taking these considerations into account when setting their strategies and building fast data systems. For example, more than twice as many preferred a commercially-enhanced open source data foundation to any other option, and more than three times as many said they’d deploy a hybrid cloud/on-premises solution to meet various needs.
These firms are heading in the right direction, and many are looking to join them if the 87 percent of organizations indicating they will re-evaluate or adopt their fast data strategy in the next two years is any indication.
There are still decisive advantages to be seized by enterprises that move quickly and intelligently to leverage fast data. “Fast data matters,” the report states. “It is fuel for real-time intelligence (and artificial intelligence) for next-gen digital business processes and customer experiences.”
Fast data is different from traditional data analytics that generate reports and dashboards. It is used to power real-time insights and decisions. It enables applying machine learning to automate applications for providing more insight in shorter time and taking action at greater speed. Businesses that can operate smarter at greater speed can gain a competitive advantage in engaging customers, optimizing operations and minimizing threats.
The potential for fast data is recognized, but not fully achieved by many companies today. Companies that want to succeed at using the growing volume and velocity of data are reviewing their fast data strategies and infrastructures to better meet their business objectives.
For that reason, fast data solutions are pointing the way to the future of analytics. For more insights from Forrester, read the full study today.
Learn more about IBM’s fast data solutions including IBM Db2 Event Store and IBM Fast Data Platform by visiting our Fast Data website.