IoT All The Things | S1 E4 – Flexing Your Data Muscles with AWS IoT



what's up guys welcome to another episode of IOT all the things in this episode we will discuss how you can flex your data muscles with eight of us IOT services I am Neel Mitra senior Solutions Architect with AWS and joining me today I'm a raga polluter I'm a product manager in the AWS IOT organization and I'm responsible for one of the services we're going to cover today it's called AWS ayat events and I'm Peter McGuire chief technology officer of eco fit networks we're heavy users of AWS and we're using a lot of the IRT services and I'll be going into a demo of how we're using AWS IOT events it's awesome so if you have notice you'll see that I'm still celebrating fathers day so welcome to all the dads we wish you all a belated Happy Father's Day so in addition to celebrating or Father's Day what else we are going to do here in the next one hour we'll start with a quick overview of AWS IOT services just in case you are not familiar with it and then we will talk about IOT data services especially AWS IOT events that you mentioned about and then we'll learn from a customer Peter over here on the IOT solutions that they're building using various AWS services we'll discuss their architecture in depth and that will be followed by a demo as well and you can see some fancy hardware's on the table we'll also be telling you what they are all about sounds exciting let's get started if you are here you already the slides yeah so if you are here you already know the potential of Internet of Things don't tell this to anyone but IDC stated there will be more than a trillion dollar worldwide technology spend on IOT applications in the next few years so we have to build a lot and Lord of IOT applications together but building IOT applications can get complex because it's multi-dimensional you can see there are different layers over here right you have the age layer where you need to onboard your device's provision devices manage your devices then you have the communication layer where we talk about all the different kinds of connectivity whether it's over Wi-Fi or Bluetooth or you know Laura or sick forks or different other kind of connectivity as well and then you can just follow your data in the cloud because you want to get some business insight from your data you are not doing all this stuff for fun right so you build your data Lake you run your analytics you are on your reporting you make this data available to different consumers so the applications that you build on top of it and also even the management applications as well like how you are patching your devices over the year over-the-air how you are organizing your devices so all those different stuff next slide and thus in the last few years we had been working backwards we had been continuously listening to your feedback and we developed different services across differently we had services at the age layer we had services at the control layer and we also have services at the data layer in today's session though we'll primarily focused on the data services layer if you want to learn more about the age layer and the control services layer feel free to check our other webinars and tech talks and different sessions in YouTube okay so in terms of data services layer we have three different services the first one is IOT analytics it's a managed service that can help you process huge volume of IOT data without a lot of coding effort so what you can do you can process all this data which could be noisy which could be unstructured and you can generate meaningful insight from the data very quickly then we have IOT sidewise which is a managed service again that we build for industrial use cases so if you have lots of equipments with SCADA and PLC's and mes and you want to liberate all this data and take advantage of the scale of cloud how do you do it that's what I all to cite wised up soon and then we have IOT events which is a managed service again which makes it easy for you to easily tect operations data and do different use cases like condition monitoring and predictive maintenance that we'll dive deeper into with that as a background let's talk a little bit about eco fit Peters also tell us about yourself sure okay well I'm currently chief technology officer at you go fit I started with this group it was about eight years maybe nine years ago and I was their first engineer and it was kind of this classic startup story where we were literally in a garage right so this garage wasn't heated it was in the winter it was cool and what we started out producing was not so much this product but very similar it was for fitness member tracking so that allowed users to log on to his system when they got to the piece of fitness equipment track their workout and then have that workout delivered up to our cloud infrastructure which at the time was not AWS something totally different something totally scalable so what we found was there were a lot of big players in that area it wasn't a super hot market we did find a hot a much hotter and much more open market in the asset management of that fitness equipment so if you imagine big gym chains like 24 hour fitness you see has great jams Gold's these guys spend so much money on their fitness equipment like crazy amounts like that's the number one expenditure and they want to be able to track that equipment figure out what's going on with it they want to know when it goes down yeah instead of having to just send techs out which is expensive and they want to know what's like what sorry what is really popular right so what should they be buying they buy every maybe four or five years this next buying cycle what should they be actually buying yeah so we take all of that data we take utilization data in through really we just put sensors on the equipment so I don't know if the camera can see this but this is our sensor right now we have a new one coming out at the end of the egg that's cool these are way these are radio transmission sensors they hook up to the fit equipment by a connection it's like a serial connection with some power on it that radio transmits back to a receiver you can see the receiver on the desk right here it looks like a router it's not really a router that uses our wireless or a proprietary wireless protocol so you can have up to 500 devices connected to one gateway in Jim all delivering that utilization data continuously back to that gateway initially when we moved over to AWS this was a little bit before IOT was really even this thing right we were doing ec2 instances we're grabbing all that data send me up by API for the cloud again storing it there generating reports sending out information to high-priority customers right now since AWS has been pushing into the IOT space and making all these great tools that that we've just talked about yeah we have started to implement those tools to not only make our data streams a little bit more dependable but also a little bit more real-time so now we're able to not only just do dumps maybe route once an hour we've got that how IOT core Greengrass IOT events combination and now we can drive the state of that equipment in real time up to the cloud and then if something happens to that equipment or something happens even to our gateway in the facility the appropriate person will receive a notification so awesome a lot more power than our system used to have and I'm sure will touch on this later on there's more stuff that we'll be doing with it yeah yes so thank you you know that's that's great it's a great story like you know how you started from essentially the bare-bones setup that you guys had your iot yeah very fascinating yeah so you talked about all this different components like it looks like you were building and into in solution right you have the embedded devices you have routers on the Aged and then you have all these applications data laying on the cloud as well so can you talk a little what about our viewers about different use cases that you are also supporting like how it is helping all the different fitness industry customers in terms of predictive maintenance or you know the asset monitoring absolutely and more that well we could actually I'm wondering if we could switch over to my laptop if that's possible right now sure this is actually our landing page for our corporate site there you go and this is our portal right here so the primary delivery method for most of our users is through our portal okay and in the portal what a user will see you when they land they'll see all of their equipment organized in two locations oh right so different countries different states or regions in those countries cities and so forth right and they will see as they drill down they can take a look you know I'll take a look say at California is California is huge for us and there's a lot of fitness facilities in California yeah a lot of good people already freakness conscious they love it and there's a lot of land that you could develop behind these 24 Hour Fitness here's two 24-hour fitness facility and a user coming into this portal to go to their fitness facility and this user might be like an executive ok who's taking a look at that overall health of all of their facilities mm-hmm and getting an idea this user may be a tack as well who's looking at their the facility that they're responsible for okay and the primary thing actually that they're usually interested is facility health right so there's a chart right here that says facility health what that means is there's six there's 89 good pieces of fitness equipment good meaning that the stats is good people have been using it doesn't look like it's broken there's no indication of that yeah and then there's one in check that means that fit that probably that fitness equipment needs a new or there's a couple in check yeah that that fitness equipment needs some attention so that's a direct feed for the customer we have a couple of different customers for maintenance and for general usage and different kind of equipments also you have like like the cardio machines and any other equipments as well that's right yeah we started out with cardio machines with with this sensor yeah we have a new sensor coming out that's battery-powered hmm and that captures the rest of it which is great because that's like 50% more the market that we can capture there yeah and our customers are really interested in tracking that information that's awesome yeah yeah and all this is possible because you are now able to capture all these operational events yeah from all the equipment's using the sensors and the gateways you have yeah and that is now coming to the cloud yeah using peyote data services yes but you can quickly get inside without much coding and stuff yeah initially we were using ec2 instances then we jumped over to like a pure lambda for service now we're moving heavily into the IOT events managed services so we're moving with the times and we're seeing a lot of benefit from it too and the ability to develop quite fast that's awesome yeah so we'll dive deep into this architecture Peter but before that we'll let out ragu talk to our viewers later what is this magic sauce under this new service a new society events how it helps you have different customers that sounds good so so let me just explain real quick on what I had events is about so we launched our events as a limited preview reinvent last year and then we just went GA last month at the Chicago summit and just like I had a high-level hired event is really built as a managed service to essentially to detect and respond to events so any ayat events that you get from your equipment if it's a you know like asset monitoring data that Peter was talking about or if it is any equipment failure type of data in an industrial context you could use all those different use cases and you can meet those needs with IOT event so what the service is about is essentially it is a set of artifacts that we have built into the service within which you can create these primitives called inputs actually these are the inputs essentially these are the the telemetry data that you are sending it sending from your I o T equipment into the service into the cloud and you use that data and use what is called a construct called the conditional logic that you built with the service and then with that you can take use the conditional logic to trigger events and you know Peter you you're quite familiar with that so this is so the idea is like you build what are called and detector models that will model the the state machine of your equipment whether it's in good condition bad condition broken condition so on and so forth and what you do is you want to take an action when they are in each one of those conditions you can do that that's all built as part of the system so can customers now use the existing IOT solutions for example if they're currently using IOT core let's say yes is it possible to use their existing solutions or it would be for new workloads only now that's a good question so this is the it is meant for like so IOT code has a specific purpose where and it enables customers to connect their IOT devices to the cloud and it's big mix that it's a connectivity solution right so when you have that data there is a direct integration from services like IOT core and IOT analytics which is the other service that you mentioned yeah to kind of integrate that directly to ID event so that you can use the data that's coming through IOT core and use the conditional logic that I mentioned to trigger actions so you can certainly do that that's great yeah so the other aspect of it that I am just getting into the details real quick here is so we have as I mentioned gain ingest data from your telemetry data from wild variety of devices you know in this case think eco fit is using IOT core as their connectivity solution in them but you can also use api's directly using what we have call as a batch put API to be able to get data directly into events as well so that's another way of getting the data and using the service in that regard the another aspect of it is the service is actually stateful and it can actually manage states of your devices at scale so I think when you are talking about especially in the consumer context when you are talking about thousands all in some some situations millions of devices keeping being able to kind of automatically fire what if what are called these detectors that you can instantiate by building defining a simple detector model is a pretty powerful situation right so when you the way you do this is you define a detector model and that will like I said the models the state diagram of your state machine yeah and you evaluate the logic and your trigger actions as I mentioned but as you're doing that you publish that detector model that you've defined and then as you are getting like I was saying at scale maybe a million devices that are sending the data to this detector model those detector instances automatically get instantiated so it's a powerful way of actually managing entire fleets of devices in a very simple manner so I just want to go through the use cases real quick in terms of where what we're seeing customers use the IOT events for you know one of them is condition monitoring so in terms of when in eff both are in industrial context or in an enterprise context we have seen because customers wanting to use essentially a simple going from a simple logic like you know if it crosses a threshold you can trigger actions or something in some simple thresholds like that as well as if you want to make some time-based type of logical workflows that you want to do that's built into the system that can enable you to do for example in here I mentioned if you want to monitor equipment equipment and trigger an action or send an alert to some technician if some fault has happened in the last five minutes or something like that you could do that with with IOT event so that will eliminate for example in this particular place any false positives you don't want to just take an action with one one bad reading for example so you could do that as part of the part of the detector model very important yep the second one which you know Peter alluded to already which is the asset maintenance use case where and we are seeing customers in our monitor their assets and then if there is a need for any maintenance that you can do that that can be done and when I mention I at analytics earlier there are actually you can do more sophisticated inputs as well right so what I mean by that is you could get not only data from your IOT sensors and that the telemetry data but also you could get some more advanced data like you know if you want to for example use IOT analytics to ingest that data and give some sort of a probabilistic input into your detector model as an input and use it as part of your logic you could potentially do that as part of you know this kind of a use case where you're not only reading it from the sensors but also from applications and others other AWS services so that's kind of that nice thing about the service here yeah so what you were alluding to is they're essentially two different kinds of data right one is a telemetry data that you are capturing from sensors and actuators and then you also have the operational data so you have a way now to bring both these different kinds of data yeah and in reaching your ultimate business outcome based on what you are trying to do that's great yep exactly and then actually that that's also highlighted in this next use case where we talk about connected products so customers actually that's one of the things that you also talked about we do where your through the customers are trying to make a decision of how their products are being used and then maybe make their decision about you know what they're going to purchase in an in terms of capex investments like so it's similar to that but even in a in other situations where we have these scene this particular use cases when you have a consumer oriented product where a customer wants to see for example how their how their the device that they have sold to a consumer whether it's like a battery or something that they have might have bought sold the customers like a power wall or or something like that where they want to see how the customer customer is charging it how the customer is how the product is behaving in terms of that customers environment right they could get that data using IOT vents and then make product next generation product decisions based on that data that data that they're receiving so we are seeing some customers that are also doing these kinds of use cases and we see some of this in action today yes so after I go through this real quick here so the last one I just wanted to highlight is another interesting use case is when customer wants to build customers want to build dashboards so this is again this is not just an alerting notification but also customer wants customers want to subscribe to essentially a topic wherein they will use that information to a downstream application you know could be as simple as a web app that could be taking this data and then being able to put a monitoring dashboard for Reno in the case of in an industrial it could be a plant technician that wants to see this dashboard or an ad in a lab environment will be a lab operator that wants to see this in this dashboard so we're seeing use cases where now they're using IT Ventures of the way that they can get more relevant data on these dashboards very cool so now that you've asked a question Nilsa let me go to a a quick demo here on how IOT events work so let me quickly change screens here from from the PowerPoint to my browser so when you go to IOT events let me just quickly do that here okay I'm logging on to the console and clicking on the ioad events console so this is your screen that you will see as a user when you go to IIT events and then what you do is I have mentioned the concept of it detector model so what you do is you can create a detector model just so I'm gonna show both from the CLI as well as a console but let's go to the console real quick here so what we have given is you know to make it easier for for the developers to use the service we actually created a couple of new things apart from what we already had in terms of a blank canvas as part of the limited preview period so as part of GA what we provided is something called a demo a demo application so what I'm going to show you is the sample application that we that's actually shows the end-to-end working of the service so if I just click on that launch demo there's always good to get customer feedback on tonight so so what they did here is when I clicked on that you know create demo is that it created what I call what we call as the inputs as I mentioned before the input is you essentially a port to get the telemetry data from your devices right and then we also created the detector model right and then so they just this is just text about what this demo application is about and you have essentially your first detector model that you can use to start monitoring essentially three states here whether your device is in a charging state in a charged state or a discharging state right and you know you can read through the details about what this demo is about but you you basically have published the detector model at this point right so and you could potentially make changes to these states you want to add a new state you could do that for example all you all you have to do is do that and then add a new state and then line up arrows to that and that's you can do that and then you publish their detector model based on that I'm not gonna do that right now in the same can you add actions like what needs to be done in each of the stage can you customize that – absolutely so you go that you go to one of the states yeah and then the way you add actions is when you you know you basically create add event and then you add an action here no this yes you have options to do like SNS mqtt we are coming up with a lot more there is a concept of timers and variables it's probably a little bit out of scope for this share the session but certainly you know please feel free to look into the documentation these are two powerful constructs within the service that you could use – for a variety of use cases like heartbeat or you know essentially condition monitoring types of use cases yeah and since you have s and s you it can act as a glue to invoke lambda function if you want to read something a custom that's secure is that's a great point yes yeah you can use SNS as that mechanism as well as MQTT you know you can go to edit over and and trigger a whole bunch of other things as well that's very interesting you like one of my customer recently they asked me that question you know we have been managing all this operational data so far with the rules engine and you know we are triggering all the different lambda functions and stuff so why do we need to use another new service so now you are showing that all the state machine that you can build and drag and drop and then publish it exactly alike so much easier that's exactly right and here's a managed service and it can do this at scale and is also a stateful service so I think you have these when you when you you're making it easier for your customers like you said you know it's simple drag and drop you really have to think about and essentially any complex event Direction use case it can be accomplished with a fairly straightforward sequence of steps to be able to accomplish that in IOT way and for our developer stone since you know many of our viewers here are developers do we also offer CLI yes we do so easy way to create model using standard formats like JSON and stuff absolutely so before I get there I will take one more minute to show that that detector model is shown up here is shown up here the one that we just created yeah and what you can do is so I want to just explain the concept of a detector instance here right which is essentially when you have to publish your detector model you get instances that you can see here as you as soon as you start sending data so in this particular case I'm going to send data right here to say send data for charging this was built as part of the demo detector model and you can see as soon as I send that data you will see a detector right there and when I talk about scale all I have to do is click one more time and if I want to create one more detector and they're all essentially next by the powerwall ID I only have to do is just make a different ID and then you go you have two detectors now so what I'm trying to show is you know it's really that you define your detector model yeah and then you can scale pretty much instantaneously and then to a really large number that's what we're talking about yeah that's the power of the system so that that's that's a gist of it on the console side now how like you ask a question about how you would do it or from a from a CLI and what's the CLI power as a support looks like right yeah so essentially we have all all these things that I just showed in terms of creating an input creating a detector model sending data all of that can be done through an API alright so what I'm going to show real quick here is essentially so we have this input called create yeah so what I'm showing here is essentially a couple of inputs that I'm going to create yeah this this is a different application that essentially demonstrates your HVAC system so at when your temperature and is increasing and decreasing in an in a building how your HVAC system should essentially trigger an action of starting the your cooling system or the heating system and so on and so forth when the temperature is needs to be heated up right so what we have here is essentially we're going to create two inputs use them in a detector model and then show and I'll show how that detect model will look like on the console for now so that it can be seen visually but what we are doing is we're going to create essentially a an input that shows what an area no a floor of this building for example yeah could look like and that's what that I mean what inputs could it take is what is modeled by this particular inputs okay so we're going to do that here we're also going to do and the output will look like essentially is a JSON document I can show that real quick here so the seed input like so yeah it's a JSON document and it shows essentially what is what are the desired temperatures for this area what is the allowed error what is the range low range low range high different types of attributes in the JSON document that you're defining and you're basically creating an input based on that okay yeah now I'm gonna create one more input create input yeah what this models is essentially as an input that is temperature sensor would receive so similar to what you were saying Gateway would receive right in your case so it would be similar to that so this is our models specific sensor that is registering temperature readings right and finally what I'm going to do from the API perspective is create my detector moral again you can do this gritty like I was saying from the console or or here and this detector model is the one that shows the different states that you are at that your HVAC system could be in and I'll show that on the console just so that it's visually it can be visually seen but it's also all of the yeah all the state machine models essentially JSON document you are correct and and even the logic that you want to put inside the different water is also defined in the JSON document okay yeah and then so once you do that let me go back to my browser here real quick and what I should see here is my yeah something called area detector model this is the model I just created from my from my from the CLI okay yeah and as you can see there's nothing there here but we will create an area essentially in a second here when we send the batch put API to send the input to this detector model but before I do that let me just quickly show you this is a pretty sophisticated model of this HVAC system where you are in a start mode and then essentially in most of the times it will be an idle mode but then there would be zoom in a bit here okay so and there could be essentially what you are doing is you are going into a low temperature high temperature threshold then you the system goes into a cooling state essentially right and then if it's a similarly if it's in a low temperatures threshold is going to go into a high heating state and then once it receives inputs saying that hey my temperature currently from the temperature sensor that it is at this it is reading you know a certain temperature it will make the appropriate transition back into the idle State or the starts yeah so so given that just really quickly what I can show is food this is the API that I wanted to show to actually create my so this is triggering essentially the detector instance from the detector model correctly so once I do this you know this other JSON that is essentially is it a sending it an input saying that hey I want to create an instance for this area so I created this it is it should show oh yeah essentially no errors and so once I do that I should go back and be able to see the detector instances for my area detector model aptly named area 51 so so yeah so we drove your monitoring area 51 that should be pretty good yeah exactly so Peter did you do the same thing with all your hardware and equipments that you have like you do create all these different kind of detector models yeah well I have only actually created one detector model okay and looking at you know understanding it better now I would separate it out into a couple of different ones okay so my detector mile is pretty complex yeah because what happens with the fitness equipment is it can go from one state to the next but there's no there's no logical transition it can go from any state that finished up with just essentially showing that the state transition does also happen okay once I do essentially my temperature now I want to send a temperature message saying that that I have a believe I have it as a high temperature mode so my detector model route change state from idle here – I think I probably need to change the number but let me see yeah I probably need to change the number so what I'm what I'm need to do it's very simple so I need to basically change yeah so while you are doing that Peter we see a lot of equipments here yeah that is super cool so what kind of like hardware and like boards that you are using yeah yeah sure so that the actual hardware layer itself I had shown this this is I know we can't zoom in so yeah that's our primary sensor that goes on to what have here which is a representation of a piece of fitness equipment sure all right so this comes from matrix fitness they were kind enough to lend this to us we do a lot of development with them okay so API development and other stuff so I brought this with me so is the fitness equipment okay this is our sensor our battery-powered Center okay right I'm not demoing this today I'm just dumbing the single sensor our gateway which I have right here and this black box right here it's not black but from the camera the back is silver but from the front it's black this drives the console so I can turn some knobs here to make the bike go faster okay example right so they saw all this together will allow me to do the demo oh okay so when you install it at the customer's location it will be essentially all this equipments together that goes there right yes so pretend you know there's 150 pieces of fitness equipment at a facility okay we install one gateway which is a black box right here okay Satchel receiver facility for the facility that connects up okay they will restore installed maybe a hundred of these on a huge piece of the fitness equipment and 50 of these onto that onto the actual non-powered weight training yeah okay we do all of that with with an installation app we put in our own installation app together we distributed that to installers that works through AWS API gateway okay right so we do all of our wrestle interfaces through there yep this lamb doesn't the back limp is do what they need to do with the database in the back that builds out a facility in the database right and then there's one button on the installation app that you hit when you've done when the installers are done that goes up to the dataverse or it goes up to a lambda function lambda function tells the database okay for this facility yeah take all that information that you have in the database yeah crunch it down into an XML file and then the Gateway itself will reboot grab that XML file pull it back down to the Gateway and then it'll build out that wireless network from the information that's in that XML file okay hi CEQA right and so now the network is all built in the last thing that the Installer does is they validate that each sensor is working go to each piece of fitness equipment spin it up hmm check through the real-time feed that the date is there you know chak chak chak chak chak then you're done for insulation okay so you probably make more sense if you pull up the architecture diagram but before we go there you wanted to show one last thing yeah I think we're we're kind of done but let me just quickly show you that this is going back to the browser yeah essentially we wanted to just show that it changed state and we also got yeah it still down okay that's great yeah thanks for showing that yeah sure so if we can go back to the PowerPoint and maybe pull up the architecture diagram for eco fate go back to the PowerPoint the next one okay so yeah so you were explaining Peter about the left side how we were doing all the provisioning for all these different equipments so how does it go to like how does it get installed in the real facility like is it like technicians go there and install it or is like self provision was they install it how it works so great we're working our way towards it okay we still have trained installers who go to facilities okay they do a really good job they do all the installation I use the app they get everything configure it all the way through you can see on the architecture yeah on the left hand side yeah there's two things there's there's the installation app that's up in the upper left and the lower left hmm there's the actual green grass core that's the Gateway right now so we're running green grass core services on that gateway so they put that all together they hit the install button after they're done Bluff but it's all together again they're removing data in oh that's nice yeah and and since it's like in a zone where you have a reliable connectivity in terms of Wi-Fi and stuff it's not like outdoor I think you do not have a lot of challenge in connecting all these gateway devices and you still have a continuous connect to the edge right my point is what additional value is Greengrass yeah bringing to your architecture here sure that does not seem like that is a lot of offline conditions that you have so there will be we're expecting some offline conditions in the future okay where we're going to need edge processing for immediate action okay machine learning figuring out the state of machinery signals in from other nice suppliers so we're keeping that door open another huge benefit from the green grass core side of things is the actual deployment of the app of them the deployment and the versioning of the Landers themselves is taken care of out of the box okay without us having to cook that solution up ourself okay so that's usually helpful for us and just be a media tie-in to services that you'd get yeah from using green grass core is hugely helpful so that that path that goes from the green grass core on the left of the infrastructure if we could bring the infrastructure back up on the screen on the PowerPoint slide oh that's mine let's use the other one give me the park very good actually yeah so the path that goes down from the green grass core on the Left down to IOT core that's super easy with what we have with green grass core in the software okay so writing it was easy which is great for us yeah that's great yeah and also I think we're discussing that you have a custom protocol where green grass is also probably helping you I'm not using standard Wi-Fi and stuff yeah yeah yeah well we built we built our own wireless protocol okay which is which is a pretty impressive thing it's a challenging thing and then yeah so there's Bluetooth there's Wi-Fi there's all these things each one of these solutions has a limitation that doesn't make it a good application for what we need to do okay so because we need to be able to have data moving pretty quickly we need to have 500 or maybe even more than 500 pieces of fitness equipment yeah on a single gateway right we need back-and-forth control and we need sort of as little overhead taking by the protocol as possible yeah so we built it ourselves okay so I've had a little bit of a hand in it we have a brilliant RF engineer back on the team who's probably listening now Krauss painting oh he's awesome too and so we built that ourselves and that allows us to just do all of that stuff all right like we can move that data we've got a ton of space and left in the packets and we need to be agile with that solution we can be agile so it's giving a lot more control into how custom you can make your devices yes we've done something like a green grass solution that's correct yeah well with green grass so that allows us to process all that different motors very quickly effectively so on and so forth okay yeah right that's awesome so we talked about how you're provisioning on the Left using a mobile based authorization authentication installation so I see a lot of custom art and called neato so how are you not using like any certificate best mechanism at all well we I mean we are using a certificate based mechanism and we're talking about IOT core okay obviously because we're doing we're doing okay you know makes it extremely easy when you in the case of when you have to have a user login so that's just kind of a straight through we're using amplify shut out the app and the web front-end and then that ties right into kognito the user management with the federated identities you can do all of the permissions everything it's really really easy all kind of the box right oh that's why we like to use it in the case of the custom authentication these are kind of machine-to-machine things that are going on okay that we have to kind of it just had it kind of came up that way that we had to use custom authenticators in this way to get to the api's this was built before we had green grass mm-hmm right and this is another thing that we could probably slave to be replacing yeah with certificate authentication which is going to be more secure than our customizers as we built them as themselves there may be some security there and then there's one key driven authorized this is because it's it's going out to trusted clients okay consumers are right data right so on the other end there's the portal on the right and there's the portal that's the output another important output is here's our open API open data API and that's for system to system integration with our with our close customers and clients we just give them a key they use that we can revoke the key if we need to and in the middle it looks like you were also doing a lot of server less computing yeah to analyze the data so that it's available for reporting and consumption that's right yeah so they're all your lambda's looks like it's full of lambdas in their land of structures where we're going and taking a bunch of data spreading it out into work or lambda workers right it means lambda worker well maybe processed 50 messages it's just easy else mmm essentially right so we've got Ronde de TLS into more formatted details that can be displayed on the portal and then you can see kind of in the center there those two structures they're building like structures in the center those guys are reports details at the portal Rio we have specialized reports that customers have indicated are really interested in hmm and so we generate that data specifically for those reports in those ETLs and then each customer can have access okay so the Opera section is essentially your badge data you're aggregating generating reports on top of it and the lower flow is a real-time stream of – IOT core and events and it's available invitation that's correct okay it sounds incredible so can we like see a demo like how all these pieces that looks like more like really good piece of art what the situation there is that we have five states that appeared that a piece of fitness equipment will go through state machine right now is really complicated when I get into the demo it's gonna be much more clear yeah so what's going on but the the situation is no matter what state you're in you can jump to another day no and so that's why there are so many transitions mm-hmm probably what I'll do in the future I'm gonna show you guys there there's kind of state of our equipment like the state of our relay yeah sure there's state of the actual fitness equipment itself yeah and then there stay to the Gateway yeah and probably in the future it's a good idea separate all those oh yeah right and then interact through different means right right so on that one right so we have seen some customers wear what they would do is they will build a hierarchy of detector models maybe that's something that's applicable yeah because what like the Gateway status is kind of like top of the heart yeah skate if the date way gateway is down don't bother about anything else exact it exactly so if they feed it to each other and that method would work perfectly for us Yeah right so yeah if we can see the demo and how this whole state machine you know fits into the whole demo sure um let's do it so could we switch back to my screen please the browser yeah that's exactly it all right so I had shown the portal a little bit before this is drilled into a piece of fitness equipment before I showed you what it looked like at a facility level and indicating like the status of each piece of fitness equipment yeah this is drilled into a particular piece of fitness equipment and fortunately it's the one that we have brought with us today right so the most important part that we should be paying attention to now is like the real time state because that's what IOT events is driving for us all right so IOT events is firing off MQTT sic publishing to MQTT mm-hmm every time a new state is entered okay and at the portal we're subscribed yeah and then we're gonna change this current status under a commitment timeline I'm gonna change this current status based off of the actual real-time state of the equipment near real time sometimes it's three four seconds there's some delay in there yes but anyway as real time as we can get okay so peace do we have a piece of fitness equipment here and just gonna let's just let's just stay on the browser for now I'm gonna plug this relay into the piece of fitness we're male plug the sensor here but I'm actually I'm gonna plug it in the wrong port so I'm an installer okay I'm in the facility okay and I'm plugging it into the wrong port this happens from time to time yeah right okay okay so I plugged it into the wrong part and I say okay well it looks pretty good it looks like there's there's some lights blinking you know everything looks you know pretty decent go back to this view movie instead of Rose it for now um I said we can probably just run it on the browser okay now okay actually maybe we should switch back yeah today yeah could we switch back to the screen view yeah yeah there you go okay so I'm back here I've just plugged in my sensor here right looked into the wrong board and I plugged it in to the wrong port all right so just give me a second here sorry guys yeah so yeah give me a second here because there's a little bit of delay on the state changes okay so I'm plugging into the wrong port click and that happens often if we could if we could switch back to my screen now the browser yeah to my browser you can see here it says relay online data not okay hmm that's bad news okay so the sensors online because it's got power okay so it's on the network we know what's going on we get some state delivery to it yes but the date is no good because it's in the wrong port okay so we see that and we say okay fine then Installer sees that and then father sees that or the facility on this is that or you could fit sees that yeah and we know what to do in that case we say okay well pull it out and put it into the different slot plug it into the different slot because I pulled it out it went to relay offline now it should actually try there you go that's awesome it's pretty instantaneous so it's pretty quick what is happening from the sensor it going through the router and Gateway turning green grass to the cloud passing to IOT events and then going to the echo v dashboard mm-hmm and that is happening like in a couple of seconds yeah yeah it's happening pretty quickly yeah that's pretty so now we know because it says on the portal okay relay online data okay boom we know it's installed properly okay can we switch back to the camera view okay so now I've pretended I'm in a facility and I'm logging into my piece of fitness equipment here just generally starting alright ready for your gym run here here I go yeah just piss on the music get my tunes going yeah exactly so I can control what's going on on the console from this box right here okay so if I I'm gonna choke I'm gonna try I'm gonna bring it down to a point where it's like not actually being used here this is ya know acquisition if there's no active yep get going on my session again so there you go no active session we actually know whether or not pretty much in real time whether or not that piece of fitness woman's being used that's awesome so now I'm gonna actually start doing some exercise Stewart here we go and there it is back in the state of really online data okay active session okay so there couple of give there couple of good things there for for eco faith yeah obviously one is the health of the network whether or not we've installed it properly hmm right I can pull it off again speaking of the health of the network yeah maybe the relay was removed maybe the relay was damaged something has happened to the relay sure so I can pull it off again and you can see if you go back to the browser it says relay offline and that happen again so we've got all these states it's like it's good to know free for it for us and the other facility owner yeah the relay is offline okay so there's something wrong with the connection to the relay and the console yeah obviously they want it online because they bought the product didn't they want that data right there's also the indication that it hasn't been installed properly again we're not grabbing data in that case then there's the actual status of the machine itself whether it's being used or whether is not being used yeah and then you can build patterns off of them correct right right is this a failure pattern right is this being used for like five seconds and then nothing and then 5 seconds and then nothing right because maybe there's something wrong with the equipment yep and then we can even have one thing that our customer is very interested in I shouldn't promise anything too much right here but as a real-time map of you availa bility of equipment for the gym customers to track to see of the equipment Oh gotcha yeah that's so cool so it looks like right it will improve the customer experience so much like sometimes it happens I go to the gym some equipments are not working I have to call someone even then it might not work yeah now it's all proactive like you already know what's happening with equipment that's right you can fix it in advance that's customers if you think of it from equipment maintenance system yeah rather than scheduled maintenance once a month yeah right or you do a scheduled maintenance you don't know what's going on you go into the gym you find five pieces of equipment right yeah maybe two more visits after that to get it fixed because you need the right parts yes now you have a much better idea okay we do have to go into this job yeah we know which pieces are at fault yeah potentially down the line if we have information enough maybe we know what's wrong with the piece of fitness equipment yes so that's a huge savings just there in the equipment minutes just improves not only the customer experience it's also operational excellence for the gym now yeah I'll try one more thing just real quick I'm I'm just gonna pull the internet out of the gateway here because I want to show you guys the gateway offline okay gateway offline is using IOT advanced timer so we are using the timer you have mentioned it in there the timer the minimum value for the timer is 60 seconds right now that's an okay window for us you kind of want to give the gateway some time just in cases having failed packet delivery or something like that it's a real problem right but at some point we will see this transition over to gay way off line yeah we can just leave this for now okay that's great so you know it's how great how even this first version of this application is working like smoothly correct but but we we have this you know saying at Amazon and probably the customers also believe the same then this is the first iteration right whatever they want yes day one whatever the future are you looking for any additional enhancements or features or capabilities yes that will be integrated into this solution in the future we're gonna find some more as we go along like right away it would be great to be able to jump to sqs from my events okay that's because just if you look at our infrastructure we use sqs like so much yeah we love to be able to just jump things to queue and then have a lambda function just kind of process it that's that's one of those things that you know we certainly have that on the roadmap yeah I know they work with you to find out a requirement and certainly have that out for you guys yeah yeah yeah no and then everything else is is good I think it's up to us to be able to start building hierarchical structures of these yes yeah and really understanding how that works one of the things that we are really interested in so on the hierarchical one so one of the things that we are also doing that's coming right around the corner is the ability to send the trigger action to another diode events detector model so you can you don't have to go through like you know the hoops of essentially maybe going to other services and and to get back to IOT back just to essentially it's a loop back go back into IOT events awesome yeah that'll be also quite beneficial for you I think that'd be great yeah I think that'd be great further things we're looking into IOT analytics okay analytics to do a little bit more in-depth analysis of the patterns coming out of the pieces of equipment mm-hmm and mixing of other signals in yeah because we do have some other sensors that may be coming online with our ships and if we can pull that into analytics and if we can pull in potentially even machine learning algorithms in there the most important thing that we deliver right now as eco feticide from our reports is the state of the equipment hmm and the more accurate that gets yeah our kind the more happy our customers get yeah so mixing all that in is part of our future plan as well that's cool an IOT analytics gives you this capability to integrate device metadata with your telemetry data and now without events as well there's an integration so you get the best of bread across all these different surfaces right to generate more and more business value out of your data right what about like other capabilities like are you looking for any device management capabilities some like device defending yeah yeah having having actually finally sat down and read to AWS well architected documentation and the IFC allowance which is a good document which I recommend to everybody to read if you're working with AWS IOT um that defend sorry security is yeah be something that I want to focus on a little bit more okay right and but also excuse me that kind of fleet management that device management we do a lot of that ourselves all right through the portal through a combination of tools rack database query if there is something that aggregates everything together and an easily reviewable and controllable platform that's great for us particularly as we had you know millions of pieces of fitness equipment attached okay that's awesome so I think we covered a lot of stuff today right so to summarize we learned how even connected devices can take advantage of eight solutions like green grass and embedded devices you have you can do a lot of custom protocol conversion at the edge you can run different local computer solutions using lambda right we learned about how you can provision your devices using different methodologies it could be like the mobile based authorization authentication key base then we'll also learn about how you are taking advantage of the cloud all the server less technology is using the ETL solution sticking advantage of all the IOT managed services in terms of connectivity with IOT core and then taking advantage of IOT events and thinking in the future to also leverage IOT analytics and other services yes and machine learning we also talked about a little bit about it so the ingress will also help you with the machine learning yeah so it carries edge start yeah very cool and then finally we looked into the dashboard how the dashboard is changing instantaneously when the devices are being made offline and connected and so this is super exciting when there's a global operation that you're doing then you have customers worldwide and you're still able to do all of these things and pretty nice and seamless manner that's beyond impressive yeah we're excited so if there will be one takeaway for our viewers before we let them go what will be that from a high or even standpoint maybe so I would say you know build iteratively so I think it's always good to put your concept on how you want to model your state's on a whiteboard or a piece of paper kind of understand that it through and then make incremental improvements as you are adding more logic and more States to your detector model that's what that would be my guidance in general well I was gonna make some recommendations initially about just going out and trying out all the tools and not being too concerned about it but I think I'm gonna pull it back and say check out well architected sure this is something that I wish I had done or actually existed like a couple years ago yeah because then I would have read that documentation and I would have gone oh that's how you do all these things yeah right so check it out if you're if you're in IOT IOT lens will give you like your the information that you need to be able to use these tools and it'll actually highlight the tools to because sometimes it's hard to with all of the tools that are available sometimes it's hard to find the tool the right tools right okay that's great so we covered a lot in the session and hope you got an idea on the different ways approaches and best practices and we cannot wait to do another episode with you so should we flex our real muscles thank you thank you Peter and tank stretchers for joining us today and apologies for starting late due to technical difficulties stay with us stay tuned and see you next time thank you thank you

Be First to Comment

Leave a Reply

Your email address will not be published. Required fields are marked *