Essential tech: Experts weigh in on what’s new in data, containers & machine learning

There are a lot of voices in the tech world. We prefer to listen to the most expert among them, so we asked some of our authors to share their thoughts on what the future holds for some of the most in-demand skills. Here’s what they said should be on your radar.

There has always been a struggle for control when it comes to business intelligence. Originally, BI was generally a centralized process due to expertise and tooling required to create reports and combine data from disparate sources. As the tools become cheaper and easier to use, self-service BI became a thing. Business users were now able to meet their own needs. But was this a new age of innovation, or the wild west and a lack of governance?

It's safe to say at this point that self-service BI is here to stay. First, we have Power BI's continued growth as a tool designed for business users first and foremost. Microsoft recently redesigned the tool to fit the look and feel of Microsoft office and be even more accessible. But more interesting than that is the other big player in the space, Tableau, was recently acquired by Salesforce. This signals, a tighter and tighter integration between BI tools and business tools such as Salesforce CRM. Much like in the BYOD era of the iPhone, IT practitioners will have to adjust to a world where everyone is able to make their own analytical reports.

The big news in containers is that Kubernetes now supports Windows, so you have a choice of using Docker Swarm or Kubernetes for your container cluster. That opens up some interesting options for organizations who want to move all their apps to containers to get the portability, security and efficiency that Docker gives you. Teams are starting to classify their apps as legacy, greenfield or brownfield. Legacy apps can be migrated to Windows containers as-is, with no code changes. Greenfield apps can be written in .NET Core to run on Linux containers. And brownfield apps typically start as monoliths in Windows containers, but using the features of Docker you can easily break them into distributed applications across multiple containers.

The exciting thing about this is that you can run all those types of apps on a single cluster. You can choose between Docker Swarm or Kubernetes, join a combination of Windows and Linux servers into your cluster, and then you have a hybrid platform for running all your apps. Every app has the same set of artifacts, the same tools and the same processes to build, run and manage them. There's a huge amount of interest in learning Docker and Kubernetes to take advantage of that. I'll be teaching some of the key concepts, and the approaches you can take to break up monoliths with containers in my workshop at Pluralsight LIVE, Modernizing .NET Monoliths with Docker.

It might seem counterintuitive to use a word like “calm” for an area as dynamic and volatile as machine learning/AI—but still, the last year has been relatively stable: a year of evolution, not revolution. However, this may be “the calm before the storm.” As business leaders improve their understanding of what’s required to expect real returns from ML/AI, organizations are doubling-down on the foundations: like enterprise-level data collection, and upskilling more staff. There’s a growing understanding that for many organizations, what they need isn’t to recruit experts who can write machine learning algorithms, but instead cultivate existing business knowledge and interdisciplinary ability of knowing when and how to use them.

Of course, the major vendors and machine learning frameworks continue to attempt to one-up each other in new features and refined algorithms, and recent advancements in research continue to make their way into these products, particularly in areas like deep neural networks (DNNs) and Natural Language Processing (NLP). We’ve also seen a welcome focus on the tools to simplify the workflow of creating, testing and deploying ML models, including improvements in Microsoft’s Azure Machine Learning Studio, Amazon SageMaker, Google AutoML, and IBM Watson Studio.

What’s remarkable is how quickly Machine Learning is becoming unremarkable. What might have sounded like a gimmick a few years ago—ChatBots, image recognition, and even the cliche of self-driving cars—is quickly becoming taken for granted, without the doom-and-gloom of previous years. Even the ML-based face-aging abilities of the recent viral FaceApp caused a variety of news stories, but most concerns were about the humans behind the app and what they were doing with the data, rather than the technologies themselves. And in May 2019, a “first” —but surely not the last— machine learning-related lawsuit: a Hong Kong-based investor is currently suing a company for multi-million dollar losses incurred after choices made by an AI-based investment program.