Summary

Abhishek Shukla also contributed to the content of this blog.

Large enterprises are using cloud in a variety of ways, but fundamentally, they’re all looking for the same basic advantages: Improved efficiency and scalability. Faster and more responsive development. And the ability to deliver more agile applications and resources, wherever and whenever they’re needed. When using cloud for industrial Internet of Things (IIoT) applications, however, the path to realizing those advantages looks very different. And industrial companies (as well those supporting them, like GE and GE Ventures) have to do things a little differently.

Facebook, for example, may run one of the largest cloud infrastructures on the planet, but everything they do lives entirely in the cloud. When you’re supporting the nonstop operation and maintenance of real-world equipment and assets around the globe, the factors you have to account for—and the consequences of getting it wrong—are quite different.

At GE Ventures, we’re always on the lookout for the next groundbreaking cloud innovation that will help industrial companies operate more efficiently, reliably, and profitably. Sometimes, that involves looking at the ways the industry and open-source community are reimagining the Platform-as-a-Service (PaaS) stack. Sometimes, it’s just about listening to where industrial companies want to go with cloud, and finding new technologies and methodologies to help them get there.

What’s wrong with legacy cloud models?

The concept of cloud-scale infrastructure isn’t new, but the technology is evolving rapidly, even in just the last few years. Industrial companies (like all enterprises) are embracing DevOps models to more quickly develop and refine applications. And they’re shifting applications to microservices and container architectures to make them simpler and less expensive to deploy and scale. Today’s popular open-source cloud development platforms have served organizations well, but many companies, keen to exploit the advantages of newer software stacks and containers, are rethinking what they want from PaaS platforms.

Legacy PaaS platforms weren’t architected to construct applications using microservices—and even more recent platforms designed with microservices in mind can still struggle with stateful applications. As a result, developers have had to use virtual machines (VMs) and typically over-provision—increasing costs. Worse, if you’re a developer using a microservices architecture working on stateful applications, and you have to call IT to provision a new VM every time you make a change, you’ve effectively broken the DevOps process.  

Newer PaaS platforms look at solving the challenge of instantaneous scale-up/down by embracing the “SMACK Stack” architecture pattern.

Innovative startups like Portworx (a recent GE Ventures investment) can also help to greatly simplify running stateful containerized applications in production.

These new PaaS architectures will allow industrial companies to fully embrace the latest DevOps methods and tools as they develop the next generation of IoT applications. And they will allow industrial applications to fully exploit the benefits of containers—both in the cloud and, in the near future, at the edge as well.

A different kind of cloud

The advantages of containers are well understood, and most enterprises plan to adopt them at some point to support microservices architectures. But very few companies are actually running containerized workloads in live production environments. Just a handful of companies today—Google, Facebook, LinkedIn—have deployed these environments at scale. For industrial companies, however, container-based architectures and microservices represent the only way to approach some of their most urgent needs.

It comes down to the basic goals of industrial companies running IoT applications. Traditional enterprises using containers typically only have to worry about them in the cloud—where there are lots of resources and lots of help. At the edge, however, industrial companies don’t have vast rack-server installations—they typically need to run applications through small gateways with limited compute and storage resources. They need a cleverly-designed and fairly lightweight stack which can run the right processes, has the capability to scale up/ down, embeds security from within the containers (and not outside them), and that absolutely will not fail. To meet all of these requirements, industrial customers will need to roll out containerized applications at the edge much earlier than other enterprises.

More work is still needed to take industrial companies where they want to go with cloud-scale infrastructure.

We still need to make it easier for industrial companies to adopt containers, microservices, and DevOps models. We need novel ideas to extend industrial-grade security to containers. And we’re looking for creative approaches to containerized IoT applications that can live at the edge, and provide fast, secure, intelligence and decision-making without having access to full-scale cloud resources.

No one can say definitively where next big innovation in industrial cloud-scale infrastructure will come from. But we do know this: there are a lot of innovative startups attacking these problems, and we can’t wait to see what they do next.  

GE Ventures

GE Ventures

Scaling ideas and growing companies that advance industries and save lives.

LEARN MORE

Related Products

About the author

Mike Dolbec

Managing Director

Michael Dolbec leads investments for GE software focused on the Industrial Internet, big data, M2M and cyber security. Mike’s current investments include: Predixion, Maana, Ayasdi, and Mocana.

Mike joined the GE team from LG Electronics, where he led venture capital and open innovation activities. Prior to that, Mike led the venture capital groups at Orange, 3Com, and IBM. Earlier, he co-founded an award-winning mobile internet startup which was sold to Earthlink. Mike began his investment career at Kleiner Perkins Caufield & Byers and Greylock. His previous investments include: Juniper, Extreme, Epigram (acquired by Broadcom), Macromind (later Macromedia, acquired by Adobe), Trilogy, and Netezza. He has held operating positions in research, software engineering, sales, marketing, business development and M&A. Mike began his career at Xerox Palo Alto Research Laboratory.

Mike holds an MBA from Wharton, as well as a Master’s Degree in Artificial Intelligence and a Bachelor’s in Biology from Stanford. He lives in Palo Alto, has two grown children, and is an avid road cyclist.

Related insights