 on the ground, presented by theCUBE. Here's your host, Jeff Frick. Hey, welcome back everybody. Jeff Frick here with theCUBE. We're on the ground in San Ramon at the GE Digital headquarters, about 1,500 people making great software to not only run the GE products themselves, but now they're taking it to the next level with the industrial internet. They're Pritx Cloud that they announced about six months or so ago and really extending that now all the way down to the edge. So we wanted to get the 101. What exactly are we talking about? Let's just get right to the basics and we're really excited to have Nikhil Chauhan on director of product marketing, Pritx GE Digital. Welcome. Good to be here, Jeff. Yeah, absolutely. So let's, we've been talking a lot about cloud, a lot about Pritx, a lot about industrial internet and IoT and the buzzwords are flying all over the place. So let's just kind of break it down. What exactly is Pritx? Yeah, so Pritx is a complete edge to cloud platform. There's a platform that could be deployed at the edge of the network as well as in the cloud. So we talk about this whole edge to cloud continuum where the platform has technologies which can run at the edge of the devices in the cloud and be able to orchestrate application logic, data and control at various levels. Right, so we talk a lot about kind of horses for courses. So depending on what the application workload is, some portion of that might be executed at the device or some subset of devices logical business unit, however you define it and some set would be back up in the cloud. That is very true, Jeff. So the way we look at the whole platforming is, Pritx is obviously a horizontal platform which allows the applications to be built at the top. As a platform, it provides the ability to run this application logic, data, control where it actually makes the most sense for the business value. So you can run all these things at an individual sensor level, at a congregation of sensors and hubs and nodes. You can run it at gateways, controllers, on-premise servers, infrastructure equipment, at the edge, as well as in the cloud. So this is the whole edge to cloud continuum we're talking about which is powered by Pritx. It went, that's interesting because some people when they say edge, I might think like that last, that last, I want to say mile, millimeter, right? The actual sensor that's on the final device. But as you just kind of walk through, there's lots of edges, if you will, as you walk back away from that one individual sensor back to different collections of meaningful units. Yeah, edge to us, we believe that edge as a word is actually talking about the physical location which is allowing computing close to the source of data. So it's either a gas turbine, it's a 60,000 foot, 65 foot tall oil and gas blower preventer, it's locomotive engines which are running at 70 miles per hour. So, and then again, you have sensors, you have edge gateways, controllers, et cetera. So wide variety of devices. Okay, so a lot of people familiar with the cloud, all the benefits of the cloud, especially on the consumer side and even in enterprise technology, why is the cloud and then this combination with edge such a powerful combination for the industrial internet? Very interesting Jeff. So because I think what we've learned so far has been that we have been great admirers of cloud, we have been using Pritx as a cloud platform and we know that cloud has those, I'm sorry, economies of scale, et cetera. However, cloud only model has its own share of limitations. The limitations being cloud computing always assumes and majority of cases assumes that connectivity is always there. And what we've learned so far is that industrial applications may not have the best optimal connectivity in there. So they either run in air gap environments, it's probably due to regulatory compliance or security issues, for example. So connectivity is really a key assumption with cloud only model. So that's one. The second one is centralizing all the analytics. So once you're taking all this data and pushing it in the cloud, all we are doing is we're basically analyzing all that stuff only in cloud, which obviously adds latency, but also doesn't give the ability to run analytics at various levels which are closer to the point of control, closer to the edge. So that's number two, that's really the centralizing analytics piece. Third one is obviously latency because in mission critical applications, you have to have systems which have to be deterministic, that have to run in real time or near real time situations, and you have to have very short, small latency time periods. So those are cloud only deployment models. Now we believe that in industrial applications, you really need to have both edge and cloud working together, working in tandem. So we truly believe that, that's really the key strategic architectural choice which has to be a marriage of both of those worlds. Edge computing, working in tandem with cloud solves and addresses these limitations. So it basically talks about four use cases or addresses those four use cases, one being it reduces that system latency. So we realize that 60% of the data that is collected loses any meaningful value for analytics in just a few milliseconds. So you really have to ensure that you have a very short latency timeframe for systems and that is where edge computing can help. So it's really reducing that system latency. It's adhering to SLAs. It's adhering to regulatory compliance requirements. So in mission critical applications such as industrial ones or majority of the ones, you really have to ensure that the system latency is absolutely short and you have a deterministic way of those systems. You give an input, you have to provide an output in a particular timeframe. So that's really addressing that SLAs. It's also ensuring that you're supporting some hybrid cloud situations where it's probably needed by regulatory compliance, some geographical compliance requirements, et cetera. So that's number two. It's ensuring that you are addressing those SLAs, regulatory compliance requirements. Third is avoiding any unnecessary transfer of data. So a typical airplane collects probably 10 terabytes of data every 30 minutes of flight. So you really don't need to transfer every single bit of it to the cloud. You have to make sure that you're doing some analytics at that point of control. Think about a locomotive engine traversing at 70 miles per hour and it's sensing all the rail adhesion things, sensing the exogenous data such as weather and it's making some meaningful analysis of the data at the edge itself. You cannot wait for that to go to the cloud and do some analysis and push the brakes in time. So that's third. The final one is where you're offloading compute intensive tasks from some of these resource-constrained devices. So think about a mobile field worker who's having a mobile phone or a device or some sort of geek robotic equipment but you really need to have the ability for offloading some of these compute cycles from these devices because these are power hungry devices. Power consumption is really critical. You have to ensure that you're offloading computing from these to let's say edge gateways and still do analytics at the edge but just save those power consumption cycles. So those are pretty much for use cases for edge computing. Really makes me think of kind of from the consumer point of view that we're familiar with is our mobile phones, right? Because if you're running an app on your phone there's certain things that the app will do on the phone and there's certain things that happen back at the cloud and you don't necessarily have to have the two exactly mirror. My favorite example was the original where the Steve Jobs I think fourth generation of iPod had no display. All it would do is we shake it, you go to the next sign. This is the stupidest thing I've ever heard. Well, he was smart. Well, you don't need that at the edge. You need that back at the computer. You can manage your playlist. You can do everything else but if you're jogging, you don't need that. It was kind of the first breakthrough where you can start to break up functionality, break up logic, break up control, break up data, two different components that are all connected to really the same application but they don't all need to exactly mirror one another and that's really kind of what you're talking about. Absolutely, that's exactly what we were talking about. It's really the whole edge to cloud computing, distributed computing architecture where you can utilize all these different layers of compute and processing so that you can run that application logic. You can orchestrate the analytics up or down based on the prescribed use case and the outcome that the customer is interested in. Right, another really key piece is kind of the partner ecosystem because you guys, you have hardware and you have software in these fields but these are big complex systems. There's historic stuff that's there. There's stuff that's not general electric that needs to work together especially in the context of an economically defined, you know, I'm going to keep stealing that line. So if you could talk a little bit about kind of the partner strategy and how you deal with kind of incumbent technologies and incumbent systems and pull those in to the value benefit for the customer. Absolutely, you know, just like any other company, GE cannot do this alone. We really need a village to support us and really the number one challenge that we are seeing from our customer's point of view is that they're sitting in a heterogeneous environment. They just don't want to deal with GE equipment. They have a non-GE equipment base as well and they have to make sense of the data that's coming off of GE and non-GE equipment. So as a platform, we're trying to ensure that we bring in right set of ecosystem players so that you have a platform player and a single point of contact for the customer so that they can build these devices in a way that is standardized, which makes these machines into software defined machines. Now you have machines which are social, which are interoperable, which have the ability to autonomously connect to the industrial internet, have the ability to execute native or cloud-based apps as well as analyze the collected data and securely respond to changes in that data. We are obviously looking at open source software as well as a number of ecosystem providers to help us do edge analytics, et cetera. But again, I think the whole point of view from platform perspective is to provide that standard scalable software framework that can work across a wide variety of machines, GE or non-GE, regardless of their vendor or vintage. Okay, and final question, how do people get started on this journey, right? Like you're dealing with people that have big infrastructure, it's in place, it's been running for a long time. They've kind of bought into doing it better. Everyone needs no one planned downtime and to get more out of what they have. So how are customers kind of getting started? Where are the points of entry on this journey? Yeah, so from a perspective, predicts is obviously a platform that provides a number of services. On the edge side, it provides technology called predicts machine, which is the technology that allows providing that standard software development kit for any kind of edge device to become software defined. So it can be a sensor hub, you can have a controller which is coming from GE or non-GE, a gateway device. So you have a number of predicts ready devices which can use predicts machine stack, become software defined and try to dial back to predicts cloud. You also have a technology called predicts connectivity which provides plug and play connectivity. So customers are using that so that they really use that service to not be at the mercy of specific carrier. It's a plug and play connectivity that works anywhere across the globe. So those are pretty much two different technologies that predicts provides. We do support a number of predicts ready devices. Predix ready devices are devices which are using these technologies, dialing back to predicts cloud, doing stuff at the edge itself, doing analytics, but really talk to predicts through standard APIs, standard integration points. So those are really three different technology areas or spectrums that the customers are using today. Okay, great. And what's the best way for people to get involved? Yeah, so the best way is predicts.io is really our developer focused website, which is where you can get all these SDKs, you can get to know what are the requirements for a predicts ready devices frameworks and really connect all these diverse set of machines and talk to predicts cloud. Excellent. Well, Nikhil, thanks for taking a few minutes. I love it. You're welcome. G is a software company. If you haven't seen Owen on the commercials yet, no doubt about it. I'm Jeff Frick. We're at G Digital, San Ramon. You're watching the Cube. Catch you next time.