What to think about when planning your Edge strategy

While I was at the Open Source Summit this summer I interviewed a variety of people on a variety of topics.  One such person was Alan Clark from SUSE's office of the CTO.  Alan shared his perspective on the Edge and how to think about it.  Take a listen (below I have included links to other interviews from the summit as well as the transcript of my conversation with Alan)

Other videos from the Open Source summit

• GitLab, an overview
• Getting started with DevOps, GitOps and an intro to OpenShift

Transcript: Alan Clark of SUSE

2022.06.22 -- 8mins

[00:00:00] Barton: Right coming to you live here in Austin, Texas from the open source summit. I'm here with Alan Clark, Alan, how are you?

Alan Clark: I'm doing great.

[00:00:09] Barton: you are in the office of the CTO and you think big thoughts, to start with, if people are thinking about the edge, it's a huge topic obviously, but what are some of the things at a high level people should look out for or start thinking about?

[00:00:26] Alan Clark: Well, you need to consider edge to be part of your regular security domain. The world has changed. We used to all live behind the firewall, right. And depend on that firewall. That's not the case anymore. So when you're thinking about the architecture, particularly as you're extending the edge out, think about that as being inclusive as part of your security.

[00:00:48] Barton:  Speaking of building it out, there's the connections and the standards seem to be, have they gotten any better?  Because there seem to be so many different standards.

[00:00:58] Alan Clark: There's a lot of standards it's [00:01:00] still evolving and that's because edge itself is, is evolving and so those standards are evolving as well as the technology changes, right? So the technology, the advancement of new technologies and the potential are pushing that edge out. And so it means we need new standards to evolve to meet that.

[00:01:23] Barton: And if you're talking about the architecture of the entire chain. How do you, how do you cut it? What are the things that you focus on?

[00:01:33] Alan Clark: one of the things I really, really push is you want consistency cuz you can drown, right?  So you're pushing that edge cuz you're trying to get to the data. and it'd be very easy to do just a one off solution out there to grab that data, but then owning and managing that forever kills you. So you want that consistency. If you can get consistency clear from the core out [00:02:00] to that far edge, you're gonna be much better off in the long run.

[00:02:03] Barton: So where do you think sensors fit? When you talk about the edge, are you considering it sensors or is that is that on beyond

[00:02:11] Alan Clark:, so yeah, they typically call the sensors IOT, but those sensors are connected to something. So you can have a tiny device, but we usually consider the edge when you've got something that's got compute power on it. So you could have those sensors plugged into something that's gathering that data and doing compute on 'em that's when it becomes very interesting.

[00:02:33] Barton: So if we talk about the edge, then we talk about rancher and K3S for example. What is K3S

[00:02:41] Alan Clark: so K3S, when you get out to those smaller devices, they just can't handle stuffing everything big on there.  So you need to tailor the software that, that is being put on that, right? Cuz that hardware can be very tiny. So we've taken Kubernetes and traditionally everybody [00:03:00] calls that K8s. And we cut it down to size to fit and tailor it to those edge devices and we call it K3S

[00:03:07] Barton: And then what was the big news that it was some company that you all of will be using your some retail company that was gonna be using all your K3S

[00:03:17] Alan Clark: there's a bunch so I don't know which announcement , you've seen, but there's a lot of companies.  We’ve , got folks in manufacturing, they're working with dev,  hardware devices out there in the manufacturing floor that are worth tens of thousands of dollars. You can't just rip and replace those. They might be 25, 30 years old, but you're not gonna rip and replace. So you adapt and put sensors and so forth on there so they can make their SLAs. So manufacturing, telecommunications, finance, medical, a lot of medical devices, so forth. So there's a lot of industries that are adopting [00:04:00] this edge strategy

[00:04:00] Barton:.  So back to the IOT versus edge, it seems when I was looking at this about five years ago, it was all about IOT. And edge was just a component just like core might be, or on the back end.  Is IOT as a concept going away and being surpassed by edge? Is the focus there now?.

[00:04:23] Alan Clark: Well the change has been in the development models.  It used to be that out at the edge was this special groups that, like I said, sitting off to the side and what they wanted to do is bring that consistency and build that into their whole architecture, right. Particularly as we want to get into real time and real time type services, which intends to mean make it intelligent.

And so typically when you're building those services, you're not just building a little point solution out there on the edge anymore. You're building a solution that's going to do some real time analysis plus [00:05:00] ship results back up to that core, which means I'm breaking up that service typically today called microservices.  But I'm breaking that service up across the layers all the way from the edge, back to the datacenter

[00:05:13] Barton: Okay. If we pivot a little bit to multi-cloud. If somebody was to come to and said, Hey, what is this? Multi-cloud and, and how would I use it? What should I think about?

[00:05:23] Alan Clark: So when I think about multi-cloud, what I'm really saying is I wanna run that service or portions of service, where it makes the most sense, right.   And that might be different clouds. It might be because they're geographically located, you know, co-located, and it could be that the services operate better in that environment, depending on what they're after.  So again, I want to be able to build those services and run them wherever I want them to be.

And that's actually one of the strengths that SUSE has is in our ability to take help you build those [00:06:00] microservices and put them wherever they want all the way up from. A data center to a cloud to, you know we could even talk architectures, right. All the way from the old mainframes down to those raspberry pis sitting out there on the edge.  And you want those services to be optimized and run to that particular environment

[00:06:21] Barton:. So what do you see as the bottleneck or the sticking point with this, would it be connecting between the various clouds? Would it be connecting on-prem would it be controlling and  orchestrating the various workloads on these clouds?

[00:06:37] Alan Clark: Well, so the key here is what I call data gravity, right? data is the gravity. And so you want to be able to work and operate where that data is, cuz moving data can be expensive so you want to be able to. once I've analyzed it. Then of course, I want to be able to consolidate it, [00:07:00] parse it and do other things with it.  Then you start talking about network, bandwidths, time synchronizations and all those kinds of issues. That's where we go back to our discussion on standards.

[00:07:13] Barton: Cool. And so just to end with, what types of things would you say is something people should put a watch on. Going forward in the industry.  

[00:07:26] Alan Clark: I would say, just watch how exciting it is because we're gonna see the demand.  Our consers are getting intelligent and they want intelligent services. Think about yourself, right? We want things fast. We want 'em now we want 'em touchless. We want instant response. And I think that's going to transform our world.

[00:07:49] Barton: cool. Alan Clark. Thank you so much. 

[00:07:52] Alan Clark: Hey, thank you. It was fun.