PurePerformance - 048 101 Series: IoT with Harald Zeitlhofer

Episode Date: November 6, 2017

Most of us remember the DDOS attack last year executed through thousands of Security Camera IoT devices. This raised security questions around IoT but also helped the public to understand that IoT (Internet of Things) is a real thing.In this session, we learn from Harald Zeitlhofer ( https://twitter.com/HZeitlhofer ) why he rather likes to call this hot trend IoE (Internet of Everything), what the key use cases of IoE are and how proper monitoring of these devices might have been the key to detect the attack before it actually happened.To learn more about this exciting next big thing we suggest to start with Harald’s latest blog posts on his most favorite topic.

It's time for Pure Performance. Get your stopwatches ready. It's time for Pure Performance with Andy Grabner and Brian Wilson. Hello, everybody, and welcome to Pure Performance. Before we go on, I have to admit that I do not have my stopwatch ready. I don't think, Andy, that we've ever acknowledged that stopwatch from the beginning and it just dawned on me. So that's all the banter I have for today. Yeah, but I actually have a stopwatch here from the PerfBytes guys.

Ah, yes. Hi, PerfBytes guys. Ah, yes. Hi, PerfBytes guys. Hey. And actually, you know what? Today's episode is something where we can draw a link from us to PerfBytes. Really? Because I think so. Because I think Mark Tomlinson has been the only guest we had on the show twice, as far

as I remember. And today's guest, today's guest speaker, it's his second time too. And before I introduce him, I do have to clarify. Sometimes people might be like, oh, no, no. There's been a lot of people that have been on twice. But a lot of those people, we've recorded very long sessions. So we made them two sessions. So technically, you can say they're on twice. But that was all in like one subject session. very long session so we we made them two sessions so technically you can say they're on twice but

that was all in like one subject session this is someone who's been on then they went away and then they came back so yes please introduce the second of two people who've ever had that honor but then again anybody else who's been on who's been who's listening we'd love to have you back this is not just exclusive or nothing but we just just – Exactly. Yeah. So today, first of all, today's topic is IoT, Internet of Things. And the expert on that particular topic from the Innovation Lab in Linz is Harald. Harald Zettelhofer. Are you there with us? Yeah, I'm here all the way from Austria.

Here we go. Hi, Andy. Hi, Brian. Woo, woo, woo. Isn't that it, Andy? The fake Austrian hello. Yeah, Andy. Hi, Brian. Woo, woo, woo. Isn't that it, Andy? The fake Austrian hello. Yeah, exactly. And Harald, if people don't know him,

I just saw some of your pictures, Harald, from a couple of days ago. You are a cave explorer. What? Spelunker? Yes, I am. Spelunker. Yes. Right. Interesting. For quite a while already. Yeah. So.

And I'm just fascinated. I mean, you have not only a gift for that, but also a gift for photography and the pictures that you always take, whether it's, you know, on top of the world or underneath, are quite spectacular. And especially those from last weekend where you found some special gems. And I'm not sure what the correct pronunciation or the correct names are for that but uh it was it was really nice you could say like that yeah so we we went to an underground mine and and really found some some uh yeah treasures there really really beautiful uh things underground and as i always try to do that so not only as I do it in my job

but also as I want to do it in the in that hobby more or less documented as good as possible and that's why I like photograph so much and that's what I did here yeah and so you're you're showing your friends obviously a different world through your lens of your camera. But now we also want to show and open up the eyes to our listeners to another new world that I think just opened up, which is the world of IoT. Wasn't it originally Internet of Stuff instead of Internet of Things? But they didn't like stuff. I think stuff sounds more fun.

Sorry, I just had to say that. Yeah. Internet of Stuff. Anyway, yeah. Yeah, there are so many different words for that, actually. stuff i think stuff sounds more fun sorry i just had to say that yeah internet of stuff anyway yeah yeah there are so many different words for that actually so some call it just industry for zero i'll call it industry uh internet of things uh actually i turn a term that i like a lot is internet of everything as we are we are not just talking about a new internet. So we don't have an internet for humans and an internet of things here.

We are talking about a new revolution in the digital transformation that's going on for a while in other topics. And now this digital transformation has reached the industrial world and we integrate devices here into our existing internet technology and we let these devices interact with humans and share the information and work together and that's why I like the term Internet of Everything a lot. I like that too, actually. And I like your explanation even better. I like the digital transformation is coming to industries that haven't seen the digital transformation yet.

More the manufacturers or some people that build goods that we use and now they're digitalizing them. And I like that that's uh i mean the digitalization itself uh has been there quite a while even in manufacturing industry so uh computer and automation uh are actually present for for quite a while now but what we are now generating or what the world is now generating is this cyber physical systems where we not only measure or or measure something or give put something on or off but we really integrate all these machines all these things that are out there. It's not only machineries.

Into our existing applications and we integrate everything. We give direct feedback. We can use that information for simulations, visualizations. And so it's really a world that integrates these devices together with the humans in the Internet. And we're not just talking about things like your Internet-connected fridge or your Internet-connected thermostat. We're talking deep into manufacturing, mining, all the real hardcore industry stuff that's all getting connected, right? Absolutely. So it's not only manufacturing. A very big topic here in the Internet of Things or Internet of Everything is healthcare or logistics,

transportations, power supply, but also the entire world of entertainment, TVs, home automation. So everything you just can imagine is now being integrated into the existing world of the Internet. I'm just looking out of my window and I see all these cars lining up here between two lights and if they would all know what's actually going on on the first light which i can see and if they would all be better connected and talking to each other they would know that they can just bypass all these stupid lanes because there's just one car in the front who is blocking everything for no good reason so uh so it's not just a computer that's in that case collecting data from one device, maybe, but being connected. Not only collecting data, but also sending the information and setting the state on the traffic light, for instance. But it could be multiple devices and really create a smart traffic light here.

So we would not have that traffic chance. I never thought about that. I always thought about the other way around, but never thought about smart traffic lights. I always thought about sending back information to the cars to make – to drive them, to kind of direct them smarter. But smart traffic lights makes even more sense. It's an interaction so on the one hand the car could send the information uh to the uh to the back end that's processing uh and calculating the perfect traffic flow and uh cars sending information here and maybe 10 cars coming from one road uh no car coming from the other road and the

smart back end automatically knows what to do and sets the correct lights or correct state in the traffic light you know that's that's a very interesting uh topic there only because you'd have to get into some sort of machine learning to throw out another phrase on the back end there because anybody who knows anything about like you know traffic stuff is you know it's not just one light all lights a lot of lights especially in cities are linked together this way there's a flow right so if you were to change the operation of one light that would require changing the other but as you're saying if if all these cars are feeding back in data to some big uh processor on the back end that can calculate and remap that.

That's, wow. Yeah, and for sure, it's a bidirectional communication. So what you just said, Andy, so that the traffic light sets information to the car is also valid because when the traffic light is now red for road B and one car is approaching on road B, it could automatically receive the information from the traffic light and and slow down automatically yeah yeah that makes a lot of sense and also if you take it one step further i mean i love to take services like uber and lyft basically if i know where i'm going i can tell my car I'm going there. And then based on that information as well,

and having it for many, many cars, we could probably much better automate the traffic lights because we not only know that there's a thousand cars driving on this road right now, but we know where they're going to. Right. Yeah. But we don't need all that because we're going to be teleporting soon enough. Right.

That's true. So when we talk about IoT now, I think there's obviously different devices. There's like something very small and tiny that's more like a sensor. But there's also, let's say, smarter devices and that may already do some processing. I think there's some terminology for the different types of devices that we have and how they communicate with the backend. Can you fill us in a little bit about the terminology here? Yeah, this world is very complex here in what we have.

So on the one hand, we have smarter devices like if you see a car, for instance, as we just talked about traffic, a car itself could more be seen as a kind of IoT gateway. So it's really a smart device that has a lot of different sensors connected. And not only sensors that measure kind of temperature or whatever, but also information like in a car, very important GPS information, so location data. On the other hand, yeah, for sure, information about gas consumption, information about speed, information about quality and health of the car system itself, the brake system. So all kind of information is collected. There are data from the motor directly. And as we are talking about devices and machines like the motor here, the engine,

we could have even vibration sensors that measure the health and the quality of the engine. And here we are in a very important situation, especially when we have devices that measure such information like vibrations. And this is not only for an engine in a car, this could also be in a pump for water or even in a railroad to measure, for instance, the quality of the rails when the wagons are rolling over the rails. All this information could be too much to be sent over the network to the back end and being processed there. And not only because it could be too much, it's also not really necessary. Typically, we have smarter devices or smarter gateways for such measurements like the car. And in the car, we have a pre-processing of this information then so the the the logic that's deployed for that uh applications is not only running in the cloud backend but also deployed to the devices itself this is uh typically or uh

ideally an automated process so that not every device here has to be configured or updated manually imagine if you have thousands or millions of devices out there that that could not be done manually so this process has to be automated and then you have uh processing of certain code already there and the data pre-analyzed and and you just send correlated data back to the uh to the back-end platform then where it's for sure used uh to do further analysis over time uh correlations not only for for one car or one, but for the huge fleet that's out there. Interesting, because I was just thinking about, like, with all the amount of data being trafficked back, how is anything going to be on the Internet anymore?

But what it sounds like you're saying is, taking the car example, you'd almost have your little local area network there where the individual sensors would be communicating back to the car, which we already know cars have what somewhere of like millions of lines of codes in them already they'd process that data right and then the important stuff would could be sent back to the central i hadn't thought of iot in that way that's really uh no that's eye-opening thanks and do you call this i think hardly mentioned you call this edge processing? Is that what it's called, the terminology?

That would be what you consider as edge processing. So really at the edge of the system out there, we still do the processing of the logic, right? I'd like to give a quick example. I just got a Fitbit from my wife and so basically my Fitbit is sending data to my iPhone where I have my app and I believe only then so kind of my I think my iPhone becomes my edge device even though I'm sure the Fitbit does some calculation as well but then if I sync it with my iPhone I think they're sending back data to probably fitbit.com to their service because then i sometimes get emails about you know what i'm

doing you're being lazy so for instance you had too many beers last night and now the amount of calories you burned is not uh summing up so yeah exactly as you're talking about fitbit here we see we can see two very different scenarios in the iot world so on the one hand we have devices that have direct direct interaction with humans so like you have your your on your on your watch probably on your iPhone you trace your your your apps your steps your your your fitness your your exercises on the other hand we have a lot of devices out there or a lot of sensors that do not directly interact with humans but really measure something in a machine in a data from from a production line and give a direct feedback to the production line itself

so the the benefit for the user is not directly from that process but indirectly. So like you get feedback from an engine in a car and directly feedback to the – when you find problems there, you can directly feedback that into the production cycle. And the benefit for humans then is to get a better car. On the other hand, we have these directly connected devices where you have direct impact if something's not working here or if something is connecting to the wrong backends or whatever problems might occur. Now, I think the IoT stuff is amazing and has a lot of great applications. But the one thing I think we've seen being the biggest problem, and I just don't know, maybe you have some insight as to if

it's being handled, if there are ideas on how to handle it. The biggest problem by far with IoT is this idea of security. We've already seen people can hack into cars. I was looking at an article a little while ago where someone using Wi-Fi on an airplane was able to hack into at least the Wi-Fi system of the plane. They hadn't been able to breach into other areas of that, right? But as you can imagine, if you have, you know, someone's going to hack your thermostat, right? Or we even saw, what was it, a year ago now about, there was the DDoS attack based on all like the security cameras, right? What's going on in the world of security for IoT? Because I almost see that as the Achilles heel of the entire cameras, right? What's going on in the world of security for IoT?

Because I almost see that as the Achilles heel of the entire thing, right? People are plowing ahead, not caring, but we know it's just around the corner. Everything's going to get shut down if we don't. Yes, as we saw, this is now almost a year. It was October 21st last year when when we had this uh big uh ddos attacks where this army of iot devices attacked the entire world so the problem that happened there is we had we had devices out there that got infected over time but no one realized that uh so and on on october 21st all of these devices started their attacks the big problem here

is that uh these these devices out there uh as we are talking about thousands and and millions of devices they are designed and and uh built very cost effective so. So they do not allow for a lot of memory and a lot of processing power. So they mainly have to be cheap. But on the other hand, that means that sometimes we have a lack of security here typically uh we when we when we see the existing vendors for uh communication for devices for for back-end platforms there is a a huge focus on on quality so uh these these devices typically have to uh authent to the backend platforms and only get access when they are authorized to do that. And typically, PKI infrastructure, so certificates and public key is used here to authenticate these devices. On the other hand, these devices are still connected to the internet.

And if there should be a chance for an intruder to hack these devices and these devices being able, or the intruders being able to modify the code on these devices, you never know what they are able to. The more it's really important to monitor what's going on there. And even these DDoS attacks could have been detected before, uh the the code was not really active but it was already on the device and just imagine you monitor the used memory of a device and send it back to a

monitoring backend and all of a sudden you find out that uh free memory goes down on more and more devices, then you can start thinking about, hey, what's going on here? And you can have a closer look and then maybe find out, oh, my God, something is loading up code to my device, so I don't have free memory available, and there will be a problem soon. Yeah, that's a cool thought. Like free memory is one indicator or monitoring the number of outgoing connections, monitoring how many bytes are stored on the device itself. And if you take this as indicators and monitor the change of them, then you can see if, you know, something has changed,

if there's a quote-unquote regression, kind of regression. And you detect a regression, even though the regression came in through a vulnerability. That's a really cool thought. Right. But if you just monitor the outgoing connections, then you only can find out, okay, there is a problem now. Yeah, yeah, there is a problem now. If you monitor, in that case, the free memory or used memory, because the infected code was on these devices a long time before the attack started because all these devices had to be prepared, had to be set up.

So if you monitor the you monitor, uh, the, the, the memory for these devices, then you can find out much before the actual attacks, uh, would start. Yeah. And, and tied to that, you know, these devices don't get updated very often. You know, some of them are out in the field running forever without an update. Um, but even if it is like, let's say it was the security cameras. Well, if, if there's a, a patch that's pushed out there, you know, tying that, tying updates to those changes so that you can see this happened when an update did not occur. Right. You would expect maybe a change if a push was, was put onto those systems,

but being able to tie directly to when an event like a net that's a, I love this idea. I think the other big issue with security, especially on the consumer devices is that they ship with a default password right and so many people don't don't change them and i hope with things like you know hospital devices and uh airplane devices and all that they're they're putting a little bit more thought into into into some of that stuff yeah that's that's that's actually a security problem with these devices that are used by humans. I would not consider them as IoT devices when we are talking about IoT here because an IoT device is typically nothing where the user logs in directly.

So this typically has to has has other interfaces now mara you use kind of already segued over to monitoring when you when you talk with our you know customers prospects and the people in the industry and we talk about monitoring are we talking about more monitoring the devices itself in the use case you just explained to us, or is it more about monitoring the back end and making sure the data gets processed in a correct and fast way? What are the main use cases for monitoring? Well, generally speaking, monitoring is a very big challenge in the IoT world. As we are not only focused on monitoring a backend application here for an IoT environment, types of devices out there, different hardware, different firmware on it, different communication

patterns, different way of authentication. So monitoring is also as complex as this world out there. It really depends on the use case, but typically I would say it's full stack monitoring as our philosophy for other applications is as well. So we really start at the device out there. it's not what we do with mobile devices to monitor user actions because we don't have these user actions. And typically, these actions itself are also not that important that much. In an IoT device, we have other really important or interesting information that we want to get. And this is on the one hand, as I said before, memory, device health in general, power consumption,

communication, so how many requests are sent out or received from other devices or from the back end. And for sure, speed performance of this communication. And not only of the communication, but also of processes that are performed on the device. Very important is also software quality. So if we have devices out there that are updated regularly and new versions of firmware is rolled out, we could end up in situations where we have 10,000 of devices out there, 7,000 work well, but on 3,000 we have a problem and we have a permanent problem. Then it's necessary to find out why that problem is.

On the other hand, when did it start and what firmware version is that? So we have to correlate this information processes here, but because the entire system is typically too large to handle that manually. Is this, I know you're working with our engineering team at Anadres on our development kits for IoT, our agent for IoT. Is this something that is already baked into what we offer? Meaning, if you use our development kit to monitor IoT devices, then we can automatically pick up things like what's the firmware version right now, some particular things from the device, and then correlate it automatically.

And then you get information about so many people have this firmware and so many people have this firmware and we have more problems here versus here yes we can already do that so it's it's actually nothing like we do with with our one agent that we just install a piece of code on a device here because we don't know what the device is like. That's why we have a different approach here. We call it the OpenKit. So the OpenKit is a set of tools, a set of libraries, reference implementations, where we show how to integrate monitoring into custom code here. We support already different languages here so we have

reference implementations for java for c plus plus for node.js and with with these implementations of the open kit we collect these data on the devices, then send it to a Dynatrace cluster, and we are already able there to use this data for these type of correlations that you just mentioned. That's cool. And then especially with the use case that you brought earlier, you're rolling out new updates to 10,000 of your devices, and then there's a problem on 2,000,

and then you may figure out, hey, the update may have failed or something went wrong, and then you get all this information, and then you can much better react on these problems. And another thing to your question from before is we are not only monitoring these devices because the important thing or the big benefit from the full stack monitoring is to see the entire system in the full context. So on the one hand, we see the devices. On the other hand, we see what the backend platform is doing, what the health is there, where are data processed on the platform?

What kind of information is stored there? What external services are called? How does the integration into the existing business process look like? So we get a full picture from what's going on on the edge device out there all the way down to maybe the database in the in the back-end application and when we see the full full context and we have when we have all these data together and we can correlate them then it's easy to find out once we have a problem where actually the root cause is located now the i I mean, there's one IoT device that I think we are all, as Dynatracers,

very, very proud of. It's our Dynatrace UFO. And I know you've been working on this project. Are you using our Dynatrace OpenKit to monitor our UFOs? The Dynatrace open kit is integrated or implemented in the in the ufo so we can monitor what's uh what what the ufo is doing so we can see when we have firmware problems and we can directly feedback that into our development cycles here to easily react on that.

Because that's pretty cool. And not only that, I mean, it's one thing to say we are a monitoring company and we provide you a development kit for monitoring IoT devices, but we actually have one of these IoT devices. And, I mean, you and, I guess, Helmut, you guys figured out what does this actually mean to have these UFOs out there that we now have, you know, we're selling and people are using it. How can we easily upgrade them?

I think you built them very easy, and I used it, a very easy-to-use upgrade mechanism. That's pretty cool. And then also the easy to use upgrade mechanism uh that's pretty cool and then also the monitoring is in place so that's um it's pretty sweet and we use our own product for that so uh because because we know it's good and we because we know that we can get the best out of it yeah yeah i was gonna say the the more important thing not that we more important than us using our own product to monitor ourselves. The fact that we have a freaking UFO.

I mean, we can go anywhere. I know. It's just awesome. For those folks that don't know what we what this is we're talking about, just go on and use your favorite search engine and search for UFO. Search for Dynatrace UFO. There's a great video on YouTube. Don't search for Dynatrace UFO, just UFO. That's what we have.

Oh, that's the one. Your next flight with the Dynatrace UFO. Yeah, easy space shuttle. No, yeah, Dynatrace UFO or UFO. There's another. I know I think this has been done we before we announced the open kit but i still think it's a funny story one of our colleagues sarah redlinger she um she has a hamster at home and she actually has a little i guess counter how often the hamster is running in the wheel.

And then she's sending that data to Dynatrace as well, kind of analyzing sleep and awake patterns and running patterns of her hamster. And it's a great blog that came out in April. But it just came to my mind, use cases of monitoring different types of devices or getting more data and what you can do with that data. That's just phenomenal for me. Yeah, and if you think about it this way, right, if you're running a typical JVM running an app server, right, you have all these JMX metrics. You have a lot of metrics built in that you can find out a lot about how that system's running. But when you're talking about something on a car, right, you don't have a performance monitoring interface. You don't have any of this kind of feedback. So what the IoT devices are allowing you to do is kind of get that monitoring in there. And then on top of it, we're going to help you, I guess, make sure all that that's running well everything is healthy and find out where those problems are but it's you know the idea the iot and all these uh in

the manufacturing lines is is really to get the the health of those systems and the feedback and knowing when things are going to fail and and you because they don't have any of this built in their physical components right they're not they're not jvms or something so yeah um harold are there any other things uh that you came across that you want to mention uh in this like session on one-on-one on iot use cases problems that people run into actually the iot world is so complex i think we could easily do two more of these sessions uh if we if we uh go more in detail with certain use cases uh but i think for now we have done a very good overview on on what's going on here what are the challenges and how to get control and manage these challenges by doing

proper monitoring well so shall we summon the summer raider let's. So shall we summon the Summaryator? Let's do it. Yes, we summon the Summaryator. So, I mean, what I learned today is that IoT should actually really be called IOE, Internet of Everything, because really what it is, it is connecting whether it's humans or devices, whether it's your car, whether it's your Fitbit or whether it is your production lane to the internet and supporting what we do in our day-to-day life. And I think we also came up with some new cool use cases that in case somebody from city planning and city management is listening, Smart traffic lights is a cool idea, I think.

And I think what I also learned is that monitoring is a big challenge, especially because there's so many different types of devices out there, different use cases. But the good news is that monitoring vendors like Dynatrace you know we think about this and we provide an open kit to make sure that we can monitor your devices and for different use cases one use case we also discussed which is hot in the news all the time is security so maybe we could have avoided the big DDoS attack that happened last year by following what Harald said is monitoring the memory usage of these devices

and preemptively would have detected that something is wrong there. So if there's anything else, I mean, Harald you said we could probably talk we could fill two more sessions on this. I'm very happy. We are very happy to have you back.

Maybe I'll be the one that does three sessions, right? Yeah, exactly. Oh, that's it. He wants to be in the lead. He wants to beat Mark Tomlinson. It'll be like back in the, there was on Saturday Night Live, there was like Chevy Chase,

who was always a host. And then Steve Martin was a host a lot. And there was this rivalry going on. And now Steve Martin was a host a lot. And there was this rivalry going on. And now Alec Baldwin, as host, I think beat everybody. So you're going to come back and be our return guest king and we'll have to find someone else to unseat you. It'll be a rivalry between you and Mark. Yeah, no, Andy, I want to concur that that memory monitoring is really a fascinating use case, and it's very parallel to the regular app world as well. If you understand how your application works before you roll it to production or even while it's in production, but if you're looking at the usage of your code through the pipeline and monitoring it through different pieces, you're going to be able to detect when something is anomalous. And same thing with the IoT devices, get to know how it runs. You know, if you know your memory shouldn't

change at all, unless there's an update, that can be a great way, as you're saying, as Harold mentioned, to detect when something might be awry. I think it's a really, really interesting use case. I also think IoT is just, you know, it's very prevalent. I think this, these, this day and age, but I still think it's very much in its infancy, right? Um, there, there's probably, you know, a million more different ways this is going to be leveraged and used in a lot of different ways in which it has to mature. Uh, so it's going to be really interesting to see where it goes and hopefully Harold can keep us abreast on all the new ways during his repeat visits as king of

the guests so Harold thank you so much for or Harry sorry I should be more informal huh Harry thank you for so much for being our guest again today any final thoughts from you final thoughts the IoT world is or the ioe world

let's call it that way it's so fascinating uh it's it's not uh sitting here and and following uh one one project uh new requests are coming in each day and and new use cases and new ideas and what could be done how it's so fascinating to be part of this development and I'm really very curious on what the future will bring here

cool excellent that's nice if people want to know more about the OpenKit, Harald, is there anything? Yes. Then please send me an email at harald.zeitelhofer, it's Z-E-I-T-L-H-O-F-E-R, at dynatrace.com. Or go to our website and you also find contact links there. If there are any use cases, any projects that you are doing and have questions on how this could be monitored or just want to talk to me about your projects, send me an email, give me a ring.

I'm happy to talk to you about that. And just to clarify, it's Harold with two A's, no O, H-A-R-A-L-D. I think U.S. is usually H-A-R-O-L-D. All different versions, usually. Oh, it'll go through, though? Or no, just the A-L-D, yes, yes, yes.

We'll put it on the website. Well, yeah, we'll put it without being a link. This way, no one does it because it'll get spammed. Also, in terms of just like... I'll find you on the website. Well, yeah, we'll put it without being a link. This way you know when he does it because it's getting spammed. Also, in terms of just like staying... I'll find you on Twitter. Yes, we'll put that up there as well. What's your Twitter handle? Same?

Hsaddlehover. Okay. And in terms of just like keeping abreast of all the cool things that are going on in IOE in general, are there any good resources you would recommend you know outside of what we're doing if people are just interested in learning more about IOE and what's going on in the industry I'm currently working on a blog post for that that could be out in a couple of days and there I will also add some links

to other further interesting readings. To be honest, right now, out of the box, I cannot provide you one. Okay, well, we'll have that link because this will air after your blog will be out, so we'll have the links to that as well. Okay. Thank you so much.

Thank you, Harry. Thank you. You're welcome thank you guys and all the best bye bye bye