Splunk Edge Hub makes physical data like vibration, power, light, air quality, etc. available for analysis and correlation with machine data. This previously uncorrelatable data makes so many potential use cases possible. I recently had a chance to chat with Tony Vincent about the particulars of Splunk Edge Hub, OTI, and applicability to healthcare including hospitals and home healthcare.

Transcript

Auto-generated captions, lightly cleaned. Speakers are not separately labeled.

Hey everybody, we’re out here at Cisco Partner Summit and we’ve been talking to Tony about EdgeHub and OTI and some of the the latest advancements. Uh what do we got going on man? >> So many of you will have joined us a year ago when we started talking about Edgehub. EdgeHub is an industrial hub. So it gives us the ability to pull data from industrial sensors, real world sensors, temperature, humidity, and vibration and carbon dioxide, water under the floor, all of those kinds of things. Knows how to talk a variety of industrial protocols, MQTT, Modbus, OPCUA, SNMP, and even can run Docker uh containers now. So that gives us the ability to run things like uh node red or litmus to talk to other industrial protocols that we haven’t gotten to yet. Also gives us though the ability to run things like uh cyber vision or thousand eyes agents uh Zeke uh camera sound and video analysis and so forth. Now Edgehub in the end ships data to Splunk. So if we segue, we can go look and see how all of this comes to life in Splunk. You want to do that? >> Yeah, let’s go take a look. >> Okay, cool. >> All right. So what have we got here? So this is OTI or operational technology intelligence at the high level. This gives us a mechanism to do fleet management. I can manage hundreds or thousands of devices the same slightly differently and so forth. I can manage both sensor technology. So I can turn on or off sensors. is I can enable built-in anomaly detection. I can also build and manage uh uh AI models that I build through what you used to know as MLTK, now the AI toolkit. I can build a model, test it, operationalize it by pushing it into the all or some or or just one of the fleet. And that gives me the ability to control where I do those AI workloads out at the edge. And remember that Edgehub is IP66 rated. So this gives us the ability to be in really rugged environments, dust infiltration, high-speed water jet, uh even industrial temperatures in the range of 140 Fahrenheit or so. So, and because it’s passively cooled, it doesn’t need to be in conditioned space. So manufacturing plant floors, places like that that are not particularly conducive to big boy computers. >> All right. Now, the fact that I can run containers is really powerful. You know, we talked about a few of those, but you could build your own as well. And because there’s an SDK, you can subscribe to data that gets to EdgeHub, but you can also push things back out. So you could, for example, listen with a specialized microphone to something and react to that and push that data, ride the plumbing that Edgehub already has up into Splunk and get value out of it in Splunk. Now we ship with a variety of builtin uh dashboards to show how the box is performing. um uh built-in sensor technology and so forth, but we also have a whole series of templates that give you kind of thought ideas that you can springboard off of. Now, these are all standard dashboard studio dashboards and that gives us the ability to really like think about how do you light things up? Now, as by now, many of you have seen other things across this back wall here at the partner hub, we’ve got a theme here to look through healthcare. And so, let’s look at a variety of ways in which Splunk and OTI and Edgehub, even Moroi sensors and other Cisco technology can be brought to bear inside of the Splunk context. So, let’s start with a hospital wing. At the nurses station, you might have a beautiful display like this that shows me all 24 rooms that I’ve got, who has issues, and whether or not I need to dispatch somebody to check on someone. From there, I might springboard into a dashboard that tells me about is the bed occupied or how’s the lighting. Even temperature in the room can be really interesting. It’s nice to know what the temperature in the room is. Why? Because if the patient is saying, “I’m really feeling cold, but it’s 86 degrees Fahrenheit in the room,” we probably have a problem that’s medical rather than environmental. And so, the ability to understand which devices are in the room, how they’re behaving, do we have, you know, there’s a typically a bed pressure plate, so we know that somebody’s in the bed. And if you think about, you know, there are lots of things in the health care space that aren’t just a hospital, for example, maybe you organize a age in place community. I might have thousands of people. We check on some percentage of those every day. And as I click in and take a look at one of those people, in this case, we’re looking at Mary, right? Mary, we might have data that comes from a variety of smart rings or smart watches or even being recorded by a health care professional that stops by once a day or twice a day and checks in. But notice that we in this case, we’ve got a series of motion detectors that tell us that, hey, they’ve been in the bedroom, they’ve been in the bathroom, they’ve been in the kitchen, they’ve sat in the living room. This tells us they’re alive. They’re moving around. They’re likely healthy. Temperature, noise level, CO2, all of those information bits help understand when the health care practitioner comes and reads the care plan notes on what’s supposed to be happening, how they’re doing, what should I expect when I see them, and they can record those same sets of notes. Now, I can take that even farther and look down into smart appliances, right? Have they used the dishwasher? Have they made coffee or tea? Have they cooked food? Again, environmental things like CO2 and so forth. Now, look, none of that makes any sense if I can’t put it in the context of what runs in the backend to run all of this. So, all of those things need to talk to some data environment. So maybe I have a data center here. I’m collecting information from temperature on racks and so forth. And I can click on one of those and take a look at an individual rack temperature at various heights into the rack. I can understand, hey, look down at the bottom. Maybe that’s the database server. I can click in and understand. Yep, sure enough, it’s the database server. From there I could click into enterprise security or so forth and understand what uh oh I don’t know maybe somebody is downloading the entire patient database >> and therefore the CPU is running really hot. The ability to stitch that whole bit together is something that only Splunk can provide because we’re the only ones that can get access to all of that breath of data. said something interesting early on that the Edgehub is running a container. Does that mean that it can run other containers? Is the Edgehub OS separate from the hardware? How is that all working? >> Yeah, so the the device itself is uh available through a select group of partners today. Um that gives the ability for for those partners to build solutions around that. But if there are solution components that they need to be able to talk to something on the outbound or to listen to something that we haven’t got a protocol adapter yet for, that’s where you would run an ARMbased docker container in the edgehub because it’s running an embedded Linux version. Right? So, so edgehub isn’t running in a container rather the inverse. It is the host and it can run other containers. There might be Zeke for packet sniffing or again thousand eyes for network connectivity and and uh uh access to to whatever um uh even even um node red or litmus if I need to talk to I don’t know profanet or s7 or some other protocol that we haven’t yet gotten to in edgehub natively. >> And then you said something about specialized microphones. So could you use all of this data that Splunk has access to, use the AI toolkit to kind of build a model and then push the inference of that model down to edge? >> Exactly. Correct. Right. And so the beauty in that is because I can get that way down close to the data. If I build a container on the reaction side, I can do that whole thing inside the device, which means if I lose connectivity, it still continues to do the thing and then backports that data to Splunk once the connection gets reinstantiated. >> Yeah, that’s really cool. So, OTI, where can I go learn some more? Where can I get my hands on it? >> So, OTI is available to download out of Splunk Base. Um, so go download Splunk B the the Splunk Base app. Remember, it gives you a whole series of these same dashboards that you’ve seen are built in. And so you can go in and they even have sample data. Even if you don’t have an edge hub, you can start to play around with them and understand how might I get value out of this. >> That’s really an incredible demo. Lots of innovation going on here with the edge devices and the data. Thanks so much, Tony. Appreciate it. Pleasure. Thanks so much.