Expert Insights: Discussing Data at the Edge with Ross Hamilton

In this edition of Expert Insights we sat down with Ross Hamilton , Solutions Architect at OnLogic, to talk about the Axial AC101.

Hi Ross! Let’s jump right into it. What has been your role in the development of the Axial Series of edge servers??

I’ve been involved since the inception of the Axial AC101, working on the architectural aspects and engaging closely with the engineering team on the development of our growing line of edge servers. I came to OnLogic with previous experience as a system lead for a lot of server-based products for the data center, so I was able to jump into the edge server project and provide the architectural guidance and direction for the overall platform.??

How would you describe the AC101?

The AC101 is a unique edge server. It combines the best of both worlds - the features that you would expect from a solid data center server (reliability, uptime, manageability), with the physical features of a device that is living at the edge (a smaller footprint and lower cost). The idea here is that our edge servers reside on premises, where latency and security matters most, scattered throughout different remote sites and geographies at the edge.?

Can you give us an example of a use case for this kind of system?

Let’s say I operate gas stations, or fast food restaurants. I want an edge server that sits in the office or back closet of that facility to perform real time AI tasks or data aggregation. Now, this can scale out substantially, so let's theorize that I have one, two, or three of these in potentially every gas station or fast food restaurant across the country. At this scale, businesses require an ultra-reliable solution that they can set up at a remote office or facility and forget about it. Our Axial Edge Server line provides a reliable compute device with a right-sized level of performance.

Where do you see users getting the most value out of the Axial line?

One area where this platform is particularly good is cloud repatriation. As you are starting to build out more features, functions, and capabilities at the edge, you don’t want to keep taking and shoving all of that data up into the cloud. It costs money to store it there, but it also costs money to transport the data to the cloud, and also to perform any meaningful computation on it. You can do a lot of that data processing at the edge with a platform like the Axial AC101, and then only output the meaningful insights. In these cases you're then doing the aggregation and reduction of the data right there at the edge.?

Where could you find an AC101 in the wild? What are some other customer application examples?

Another example of where a system like this can be found is in a manufacturing facility. There's this whole notion of the edge where the compute is located near where the data is being collected. Oftentimes, we see edge computing devices bolted onto the base of a robot, and it’s controlling a robotic arm, or it’s next to a production line doing some sort of sensor data collection in a production process.?

This is where a lot of our rugged and industrial solutions perform very well, such as the Karbon 800 Series, the Karbon 400 Series, and some of the Helix Line. These devices are continuously collecting? a lot of data and this creates a need to process all of that information on an edge server. This server might sit in a back office or closet, adjacent to networking equipment, within the manufacturing facility. That’s where the AC101 can shine.

Can you talk a bit about trends in the edge server space broadly - and what AC101 represents??

It’s an interesting new market that is being explored on a lot of different fronts. There are companies that have started to move up the stack, like OnLogic, offering more horsepower and higher performance computation, moving towards the server class of systems.?

With an industrial computer based application, we are generally limited by the amount of expansion features, processing cores, and memory that can fit within the small form factor. These systems are leveraging historically consumer grade processors such as Intel? Core?, but through our strategic partnerships, we are able to offer and support long life embedded options. Where edge servers are going is hitting that upper boundary of what can be done in the Core-based processor space. Soon we will be reaching into the typical data center processor SKUs for those workloads that really require the next tiers of performance and I/O.?

At OnLogic, we're focused on the on-premises higher performance compute. With that comes some unique challenges around size, shape, features, functionality, acoustics, and power. These are all things that are related to the customer's environment and the capabilities that they require. For example, if a customer needs an edge server to be mobile, or to live in a converted custodial closet - in both scenarios, they will need a computer that can function properly, and reliably, in that environment.?

And what would you need to be conscious of in a custodial-closet-turned-data-center?

A big one is lack of airflow. A closet is not designed to have a whole rack of servers in there from a thermal and power perspective. Data center servers are drawing significant amounts of power. When you are at the edge, that becomes a boundary condition. You are limited in terms of power availability. You also have some security ramifications. For example, how do you ensure that the data on the system is secure? How can you make sure that no one can tamper with the edge server? As we continue to evolve the edge server space, these will be things that will become increasingly important. ?

Finally - what do you find cool about our new edge servers?

I think the Axial Line of edge servers poses a lot of unique opportunities for additional AI processing at the edge. The support for a powerful discrete GPU at the edge opens the door for a lot of AI processing, including Machine Learning (ML), robotics and more.? It holds a lot of promise and prowess as we start to see more ubiquity of AI in the world, and at the edge.?