ITechnology Interview with Dominik Pilat, Field CTO at Hivecell

“Edge computing has introduced more sophisticated compute to the edge that is distributed, but always comes with some sort of a centralized management.”

Hi, please tell us about your journey into technology space and how you started at Hivecell?

Growing up in Germany, I always knew I wanted to be an inventor. As a result, I studied physics in school, earned both undergraduate and doctorate degrees in the field, and invented novel measurement instruments. However, since the real-world impact of scientific innovation can often be quite delayed, I aimed for a career in the private industrial sector. One such company I applied to was SVA System Vertrieb Alexander GmbH, who wanted to get into IoT but really didn’t know how or what this truly meant. So they asked me to lead the charge and I ended up building a large IOT team at a German systems integrator over the next five years. I was extremely excited about this entry into the world of IOT and edge computing. Now at Hivecell, I am once again at the forefront; this time with edge-as-a-service development and I am just as excited. My mission at Hivecell is to help companies globally extend their applications to the edge while developing unique solutions to better their business.

Can you define Edge-as-a-Service (EaaS) for our readers? How do these differ from IaaS and PaaS offerings in the modern cloud ecosystem?

The “as-a-service” distinction simply means that the company you have hired to handle your edge computing is taking care of the process from start to finish. Period. As a customer, you don’t own or need to worry about the hardware, software, or maintenance required to operate the edge technology, which is great. Let the experts handle this. Hivecell, for example, sends our customers the edge computing nodes needed, which are super simple to deploy. Anyone can easily create a cluster at the edge. After the setup, Hivecell handles everything else from this point on—including the hardware, the operating system, and the platform itself. Our customers only need to consume the platform and are responsible for their applications.

Read More: ITechnology Interview with Neal Gottsacker, Chief Product Officer at Nintex

Tell us more about the edge computing trends and how these have changed in the last 2-3 years, especially during the pandemic?

Companies will often say they have been using edge computing for years, even decades. And they are probably right. They have been utilizing little computers called PLCs or microcontrollers. However, these are very static by nature, decentralized, and in some ways isolated. And, even though they are extremely reliable, they are also very limited in their capabilities. Edge computing has introduced more sophisticated compute to the edge that is distributed, but always comes with some sort of a centralized management. Both the compute and the management have evolved over the years: In the beginning, companies only offered simple arithmetics as edge computing, then they slowly allowed scripting and programming. Today, it is possible to use the same deployment methods on edge as in the cloud, like docker and kubernetes, making any type of compute possible and economically feasible. The possibilities to deploy GPUs at the edge also pathed the way for machine learning and, in particular, neural networks to be utilized at the edge. On the management side, the platforms additionally became more sophisticated, allowing companies to use the same deployment techniques they use in the cloud to be used at the edge at scale, e.g. depoying helm charts to thousands of edge compute clusters. This means all together. Today, we find the new edge compute coexisting with the old one—a PLC next to a Hivecell cluster, the PLC controlling an assembly line, and the Hivecell cluster running a neural network that utilizes PLC and camera data in order to control product quality.

How do you use AI and machine learning in your product development? Any recent upgrades that have been designed specifically with AI ML in mind for Edge as a service?

At Hivecell, we have had AI and ML top of mind from the very beginning stages of product development. We have optimized the hardware for AI by utilizing Nvidia Jetson technology. And we have chosen software platforms, like Kubernetes and Kafka, that data scientists and data engineers use to build their architectures and, in particular, run their inference. Our recent partnership with DataRobot, who delivers an ML-OPS / AI cloud platform, is a prime example. This alliance enables enterprise organizations to solve bigger challenges at the edge by deploying machine learning models on-site and outside of the data closet.

Please tell us about the biggest challenges and opportunities you met in 2021?

The edge is messy. Every edge environment is different—different networks, different physical conditions, different skill sets on location, etc. This might also be true for data centers, however, the number of data centers per company is usually one or in the low single digits. In comparison, our customers have hundreds to thousands of edge sites making deployment, management and maintenance very challenging. I have seen what can go wrong if you try to tackle this complexity with unfit technologies, and I have seen how lengthy and tough the path can be, when companies try to solve the edge alone. However, I have also seen how Hivecell has slowly, but surely, revolutionized the edge space, giving developers the same platforms they are used to in the cloud, but deployed on edge, and taking care of everything else. This is the reason why I came to this company and this is the biggest opportunity I’ve met for edge. A true convergence between cloud and edge.

Read More: ITechnology Interview with David Karandish, Founder & CEO at Capacity

Your predictions for the year 2022—what does your advice on how to build a solid Edge-focused data roadmap for the coming year?

As the global edge computing market is expected to see massive growth in the coming years, it will become more and more important to change current mindsets of what we used to know in order to leverage what is to come in the future. Currently, everyone utilizing edge computing has a background in the data center. These people will be well advised to embrace the paradigm of distributed computing in order to succeed at the edge. My advice for building a solid edge roadmap is to identify specific use cases, start building their business cases early on and then just start. Start doing edge.

Your thoughts on leveraging AI, analytics and automation for developing world-class Edge solutions?

The technology for developing world-class edge solutions exists. Now it’s time for organizations to play catch up with the latest IT evolution. Where to begin? It’s my recommendation to start with the end in mind. Imagine we have mastered edge compute by the end of next year, what are we able to do then? Then, with that as your north star, start. Be proactive, get a bloody nose, get up again and just do edge. As with every disruptive technology, there is a huge benefit to do it early on, so there is no better time to do it than now.

Read More: ITechnology Interview with Sagie Davidovich, CEO at SparkBeyond

Thank you, Dominik! That was fun and we hope to see you back on itechnologyseries.com soon.