David C King, FogHorn Systems | CUBEConversation, November 2018

Summary

In this CUBE Conversation, Jeff Frick interviews David King, CEO of FogHorn Systems, about the convergence of edge computing, fog computing, and cloud computing. King explains fog computing as an advanced form of edge computing that brings cloud functions like big data analytics to industrial environments. FogHorn's focus is on delivering AI capabilities on live-streaming sensor data to optimize industrial IoT processes in real-time, reducing the need to send massive data volumes to the cloud. The discussion also covers the integration of IT and OT, the challenges of cybersecurity in connected systems, and the potential of video and audio sensing in industrial applications.

Takeaways

• 🌐 FogHorn Systems is a company focused on fog computing, which is an extension of edge computing and aims to bring cloud computing functions closer to the source of data.
• 📈 Fog computing is designed to perform analytics and machine learning on live-streaming sensor data, reducing the need to send massive amounts of data to the cloud.
• 🤝 The convergence of Operational Technology (OT) and Information Technology (IT) is crucial for leveraging AI and IoT in industrial settings, despite the historical separation of the two domains.
• 🛠️ FogHorn's technology can run on a variety of hardware, from small devices like Raspberry Pi to larger systems, emphasizing the flexibility for different industrial needs.
• 🔒 Security is a significant concern as connecting OT systems to IT networks can introduce vulnerabilities, despite the benefits of real-time data insights.
• 💡 The industrial IoT is not just about data collection but also about applying AI and machine learning to improve operations in real-time, leading to significant economic benefits.
• 📊 FogHorn's stack is designed to handle high-frequency data from industrial machines, enabling on-the-fly computation and decision-making.
• 🔄 The concept of 'ML on ML' or machine learning models improving other machine learning models in an automated loop is a key aspect of FogHorn's approach to industrial AI.
• 🚀 FogHorn's technology can be integrated into existing industrial systems, either by sending processed data back to the cloud or directly into control systems for immediate action.
• 📹 There's a growing trend in industrial IoT towards using video, 3D imaging, and audio sensing for insights, which was traditionally underutilized.

Q & A

What is the main topic of discussion in the video?

The main topic of discussion is edge computing, fog computing, and cloud computing, with a focus on how these technologies intersect and their applications, particularly in industrial IoT.

Who is David King and what is his role in the discussion?

David King is the CEO of FogHorn Systems, a company focused on fog computing. He is in the discussion to provide insights into the company's background and the concept of fog computing.

What does fog computing represent according to the discussion?

Fog computing represents the intersection between cloud and on-premises computing, aiming to bring advanced computing capabilities like analytics, machine learning, and AI closer to the source of data, typically in industrial environments.

How does FogHorn Systems differentiate between edge computing and fog computing?

FogHorn Systems views fog computing as more than just edge computing. While edge computing has been around for decades in industrial settings, fog computing is seen as a more advanced form that applies cloud computing functions, such as big data analytics, in an industrial context or directly on a machine.

What is the significance of 'big data operating in the world's smallest footprint' mentioned by David King?

This phrase signifies the concept of performing complex data analytics and machine learning on a small scale, close to the source of data, which is essential for real-time decision making in industrial IoT without the need to send massive amounts of data to the cloud.

What are the challenges in merging OT (Operations Technology) and IT (Information Technology) as discussed in the video?

The challenges include historical separation and different priorities, such as real-time control and safety in OT versus data-driven insights in IT. There's also a need for education and understanding between the two fields, as well as addressing security concerns when connecting previously isolated systems.

How does FogHorn Systems address the issue of data persistence and analysis in industrial settings?

FogHorn Systems focuses on performing analytics and machine learning on live-streaming sensor data at the edge, reducing the need to persist large amounts of data on-premises or send it to the cloud for processing.

What is the concept of 'ML on ML' mentioned by David King?

'ML on ML' refers to the concept of machine learning models improving other machine learning models in an automated fashion, such as updating a global fleet-wide model based on insights gathered from edge devices, without human intervention.

How does FogHorn Systems handle the computational challenges at the edge, especially with limited resources?

FogHorn Systems has developed a software stack that is lightweight and OS-independent, capable of running on small form factor devices like Raspberry Pi, making it suitable for edge environments with limited power and connectivity.

What are some of the practical applications of FogHorn Systems' technology in the field?

Practical applications include condition-based monitoring, predictive maintenance, asset performance optimization, and plant-wide optimization. The technology also enables the use of video, 3D imaging, and audio sensing for insights not traditionally derived from such data.

How does FogHorn Systems ensure that its solutions are non-invasive and compatible with existing industrial infrastructure?

FogHorn Systems ensures non-invasiveness by developing solutions that can run on existing hardware, such as PLCs, and by initially providing alerting and insights without directly interfacing with control systems, allowing for gradual integration and proof of concept.