How IBM Uses Reactive for Insights and Data Science

DZone 's Guide to

How IBM Uses Reactive for Insights and Data Science

Reactive architecture stresses responsiveness, resiliency, and elasticity. So see how IBM is putting it into use, including thoughts on moving to production.

· Java Zone ·
Free Resource

Sebastian Hassinger, Developer Partners and Ecosystems at IBM shared his thoughts on the current and future state of the reactive platform in advance of the Reactive Summit this week in Austin, Texas.

From IBM's point of view, why is Reactive such an important movement around the design characteristics for enterprise applications and systems?

IBM has been helping clients manage and gain insights from the increasing volumes of data for quite some time — investments in open technologies like Spark and Hadoop and offerings from our analytics, data science, and machine learning (ML) businesses have been leading the way.

When a business reaches the maturity needed to move from building insights to operationalizing data science, the use cases bring unique challenges that the Reactive architecture is very well suited to address. The Reactive architecture stresses responsiveness, resiliency, and elasticity attributes that are necessary to build applications with the largest volumes of data moving at the highest of velocities. 

IBM obviously knows AI types of workloads — how are these general use cases around data-driven applications stressing the need for Reactive?

AI is about leveraging the insights derived from Data Science activities — bringing data together in a data lake, diving into the data with advanced analytical tools, building machine learning (ML) models that provide deep insights based on the enormous collection of data. These activities deal almost exclusively with data at rest. When moving these insights into a production environment, your ML model is no longer looking at historical data, scoring past transactions for the likelihood of fraud, it needs to score live transactions in real time, providing the insight within the window when the application can act upon it. Reactive architectures are designed for applications built around these fast data use cases. 

We're in the middle of a bit of a developer renaissance with containers, microservices, and general status quo of enterprise architecture patterns being re-thought. What advice does IBM give to enterprises that have huge investments in existing assets running on legacy application infrastructure, but are looking to participate more in the scale-out trends that are the drivers behind Reactive?

Our collaboration with Lightbend around the Reactive Platform offering will enable enterprises to create a modern application platform that can be deployed in whatever way is appropriate for where they are in their own IT investment strategy. Reactive Platform applications are "Cloud native" in the sense that they are built to handle enormous scales of data, deployment, users, and end points, whether they are initially deployed on bare metal in an enterprise's own data center or in a container on IBM's Cloud or anywhere in between.

What if any predictions could IBM offer about the popularity of Reactive? The Reactive Manifesto has 20k+ signatories, it's been translated into 40 different languages — it seems to have a lot of momentum. Is this something that could reach the level of the momentum of movements like Agile? What does IBM see as the tipping point where Reactive is a requirement not just for web-scale startups, but enterprise at large?

The earliest adopting industries are predictably those grappling with the highest volumes and velocities of data — in particular, the financial services sector. However, there are many early adopters whose use cases are driven generically by scale — massive numbers of users, massive numbers of massive numbers of IoT devices, massive streams of data. There are no trendlines that show diminishing scale as the world becomes increasingly digital in every imaginable way, and so it seems reasonable to expect that the drivers behind Reactive will affect every vertical and every market sooner or later (generally sooner).

IBM is both partnered with and invested in Lightbend. Both companies are advocating Reactive, both companies are working closely with JVM language developers, and both companies are helping enterprises modernize from monolithic applications to more scale-out, Reactive Systems. What does the future hold in store for how the two companies work together? 

We are extremely enthusiastic about the collaboration today and into the future with Lightbend. Our open technology collaboration began with the Scala Foundation and continued through the Spark Technology Center and will continue in a variety of forms in the future. From a market offering perspective, IBM is starting with a pure OEM offering of Lightbend's Reactive Platform to its clients, but work is already underway for greater integration with IBM's Cloud strategy, our broader portfolio of data science and analytics offerings, and on defining future directions for joint offerings that bring new value propositions to the market and to our shared clients. 

data science ,java ,machine learning ,reactive

Opinions expressed by DZone contributors are their own.

{{ parent.title || parent.header.title}}

{{ parent.tldr }}

{{ parent.urlSource.name }}