AI features in Catalyst Center

Thank you. So let's start team. Now.00:11Again good morning, good afternoon, Welcome to today's APJC expert insight series. The topic for today is AI features in Catalyst Center and we have two amazing presenters with us, Rahul, and Vani.00:15So Rahul will be our presenter and he has around 14 years of experience in networking across enterprise campus and service provider environment, and he is also a ccie enterprise with experience in network automation, programmability and is networks.00:32And we have our Vijay as demo specialist. He's also a customer success specialist for enterprise networking in the APJC region. He has over eleven years of comprehensive experience in routing, switching wireless, and even technologies.00:51So before we actually jump into the session, just few rules so that we can have a smooth and seamless meeting. So for the comfort of everybody, we have muted all the attendees and we will be starting the Slido in a minute.01:11For you to ask any questions that you can have, you might have and we also have the chat feature to mention if to mention any of the technical support needs that you might be having, let's say you have some audio issues, please put it in the chat. And also we have closed captioning available in the bottom left side of your screen. So if you would like to use it, please.01:27Think so.01:47And we also have the emotion emojis option in the bottom right of your screen. If you'd like to use it, please do so to share your emotions and this session is being recorded and the recording will be shared with you over the email and also will be uploaded on the cisco community.01:48Right. After the end of the session, once you end this call, you would be redirected to a brief survey which would take less than 2 min. We would really appreciate if you could share your feedback so that we could improve our future offerings.02:08So before we jump into the session, we do have a very cool tool that we have recently released at a Cisco IQ and we just like you to explain what it is in this 2 min, just like brief what it is.02:24So Cisco IQ is a simplified interface where you can manage your assets, get assessments for your devices like security one what are the security vulnerabilities that might be hitting my devices? What are the field notices, and now recently we also have added configuration best practices, security best practices.02:41And certification expiration, so these are some of the features that are available on Cisco IQ which are personalized based on your devices that you will be onboarding on Cisco IQ and all these assessments that I just mentioned are proactive. You don't need to do anything every day as per the schedule that you.03:00Who will be doing this will be, these assessments will be updated automatically on the tool and these will be most, these will be predictive. So all you need to do is just add your devices, add basically your service contracts and all the devices that are associated to the service contracts will be appearing on Cisco IQ and.03:20You'll need to send telemetry, but yeah, once that is done, you'll be getting all these assessments for free. Yeah, it is for free minute. And with that now, I would like to hand it over to Rahul to officially kick off the.03:38Over to you Rahul. Thanks a lot Lassia. Thanks everyone for joining. Good morning, good afternoon from where wherever you are joining from. So, today's agenda, I hope I'm audible and can somebody confirm? Yes, Rahul.03:54All right, so alright, so with that we'll start with the today's session, right? So the agenda of today's session is, you know, we'll, we'll start with the introduction of AI network analytics. All right, what AI network analytics is for the ones who have.04:11You know, not use this feature in the catalyst center yet. We'll do a small introduction to it. Then we'll discuss the key use cases of AI network analytics. Now we'll start with the RF optimization, right? And we'll also, you know, look at the productive troubleshooting with the help of the AI features.04:27How a pro pro productive troubleshooting of wireless network issues can be done. And we'll also discuss AI endpoint analytics at the end, you know, when we cover AI plus the proactive troubleshooting that we're talking about. All right, so that.04:46So we will start with the network analytics and see how we transfer the raw telemetry which comes from switches, controllers and access points into from our network into actionable operational insights. All right.05:02So, you know, enterprise networks produce millions of telemetry events every hour, and this is far more than any engineering team can manually analyze. Now in this architecture, say your catalyst 9800 controllers and catalyst switches.05:18A Stream de identified performance metadata up to your local catalyst center appliance which is like, you know, hosted on prem. So from there anonymized KPI metadata is securely sent over TLS to the Cisco AI cloud.05:34Right. I just want to also emphasize that NO customer payload, NO username passwords or NO internal IP addresses can ever leave your network. What happens is that the AI cloud aggregates this data, runs you know multi week machine learning models and pushes the actionable baselines and insights, right?05:49You know, back to the center dashboard. This is all about the kind of architectures we can talk about when it comes to the network analytics. All right, so then next part is, you know, what, but this basically this slide tells us that the air network analytics is powered by basically five core engines, right?06:10Basically, you know, a dynamic baselining and anomaly detection which calculates the true mathematical definition of normal for every individual building or floor. Then we have proactive RF optimization which learns usage patterns to tune channels and power without.06:31Disrupting users, transit insights, it automatically you know identifies systematic issues across your sites. Then we have comparative analytics which benchmarks your performance across buildings and against peer networks in your industry.06:48And finally we have predictive analytics and machine reasoning, which pinpoints the exact root cause of onboarding failures and prescribed this as a fix. This is domain specific AI built on decades of Cisco networking and technical experience.07:05Now we'll talk about as we were discussing earlier, right, in terms of the key use cases of your network analytics where we start with that. And the 1st use case that you know we'll, you know, discuss on is the proactive troubleshooting of wireless network issues.07:24All right. So the 1st one is the site analytics and this will discuss you know in a larger detail when it comes to demo, right? But you know, the very 1st use case is site analytics which monitors the client experience across the four, basically four.07:43You know, domain code dimensions. The 1st one is the onboarding attempts and durations, which includes the association authentication DHCP and DNS, right? Then, you know, it also covers the connection speed and data rates. It covers you know roaming performance and roaming durations.08:00And finally the RF coverage and quality, basically, you know, checks the SSI and SNR. It also automatically aggregates these KPIs down to the global sites, two areas, building floors to the individual access points.08:18Then comes the baselines, you know, so what we are talking about here is a dynamic baseline or dynamic baselining. Now what happens with the static threshold is that the settings with an alert at 5 % say authentication failure causes false alarms during morning login peaks.08:36And, you know, it mrs. a real outage late at night. While dynamic baselining machine learning calculates a expected performance band for every 10 min interval of the week and an issue is generated only when life telemetry breaks outside its predictive envelope.08:54Eliminating the false alarms, right? So, we will look at this in the demo that how you know dynamic baselining works instead of static baselining which we are used to in the traditional way of monitoring our network.09:13Then is the network heat map. Network heat map is another important feature specifically for the reporting part. What it allows you to do is to, you know, quickly stack and rank your network based on KPIs and allows you to quickly identify any product problematic device.09:31You can visually see your network up to last 30 days I identify both top and bottom performers, multiple KPIs available. You can have a look at that. Another feature is the, you know.09:49AP performance advisories, it uses machine learning over a four week window to benchmark access points.10:09Against their physical neighbors, it automatically flags hardware that is under performing, you know, such as APs operating in a low density coverage gaps or you know high interference rules, right? And then we have the AI driven issues.10:16You know, the catalyst center duplicates thousands of raw syslog and SNMP traps into high fidelity statistically driven issues. It is categorized you know clearly by business impact going from B one to P four, right? And you'll be able to see, you know, the normal.10:35Issues being raised within the center and the year driven issues, right? So the ones which are bracketed into the air driven issues, these are the ones which are being generated by AI, right? And you can click on these particular issues and you'll be able to troubleshoot.10:54And there's a machine reasoning engine also within the catalyst center that can help you, you know, troubleshoot why a issue has happened. Now this is the time for the very 1st demo. I'll you know hand over to Vijay for the 1st demo.11:11Thank you for the wonderful presentation. Now, let me go to the demo site and just a minute. So now the very 1st thing I would like to start with the persons which I'm using in my lab, that is 2379.11:29But we have a recommended release that is 2T710 as well as the 23711. Both are the suggested. So I'm using this 2379 as this one's available for me and.11:48For using the any of the AI features right in the catalyst centers, the very 1st thing I'm assuming that this system settings page, we have this AI analytics that should be enabled. If we go to the catalyst centers under the settings page, you can see we have a Cisco AI analytics where we have all the I think whatever.12:00The AI features we have it, right? It should be enabled like if we want if we have to use this. So the very 1st thing we have to do the checkbox, I mean right now it's already enabled in my cats, but you should enable all those settings, I mean the AI analytics features on this and the.12:20With the cloud is basically where you want to, which cloud basically you want to do, use it out. There are two clouds basically the US there's a Europe Germany cloud. So you can just select the cloud, which cloud is basically using it out. You can send it out and get the AI features enabled so that you can get the AI insights in the cat three so these.12:40Algorithm will be done on those clouds and all. And we also have a privacy data sheets. If you want to see what are the data sets we which which is being shared on the cloud, you can have those have those analysis on the sheets as well.13:00So now let me go to the AI network analytics page I mean, if we go to this assurance, we have this particular sections of AI network analytics. Under the AI network analytics we have couple of features you can see for instance, network heat map, peer comparison, network comparisons, right? Baselining and AI rfm. So let me.13:13Go to one of it, like 1st one the trends and insights.13:33So if you go to the trends and insights frame, you can see we have the two main features under it, the AP performance advisories and the trend deviations. So these all are the very useful insights which is getting changed on the period of time. If you see right now we are seeing on this time frame, these four insights are coming.13:38I mean from 9 October 9th October to 2026 to the 10th of ten 8th 2026. So the four insights which is being highlighted on it. So which is basically giving the information on which of the APs have a high coach and interference, high density, right? High client activity low.13:58All those are the different performance advisories which are which are highlighted by the AI cloud and giving us the information how the how the how how the wireless network is basically performing it out. So let me just pick it up one of it like low AP density. Here you can see I mean there's a low AP density.14:19With the 5 ghz back and this is the one of the building and top one radio affecting the client, meaning that one of the AP which is affecting the client experience because of the which is being found. How the AI have calculated it out, you can see at the bottom right we have two important dashlets. One is this radio.14:38In your network and second one is the radio is causing the bad bad experiences. So if we compare the reference radio, right? I mean almost all the other APs which we which we are seeing, right? Which are taking the power distribution of 19 to 20 dBm around the 0.7 percentage of time, right? But one of the AP right which is taking a.14:5819 to 20 DBM 9.5 percentage of time means it's running a high power level almost at all the duration here. If you can see this pink color, right? This is basically our one of the AP which is taking a huge power label that almost every time. So based on these comparisons that AI highlighted that there is a.15:18Low evidency at this area in this building. So these are the insights which are very very useful when we are doing wireless performance, right? When we are measuring the wireless performance, so these insights really makes a, makes the highly valuable insights and based on this data, we can improve our wireless experiences by putting one of the API.15:38It's different on the things, right? Whatever the insights are coming, we have to take actions on. 2nd important things which is coming the trend deviations here you can have a look. We also have these features where we have categorized under the capacity as well as the throughput and under the capacity you can see basically what it does.15:58It's it's going to calculate based on the capacity, how much your APs was handling previously from the past data perspective and how it's being currently doing it out. So let me just take an example of one of it. I'm just taking on the 5 ghz band, right? My client count is over the past few weeks it's changed.16:18So if we just click on it, it's going to give me a piece on graph it's loading. So here you can see, right? So weekly client load that is around the 23 for this 5 ghz on this AP and all those circles are basically nothing, these all are the AP which is being there. So.16:39Right now what it's highlighting on the week one on one of this AP, basically which is which is handling this 32 number of clients that you see at the bottom client count, right? And suddenly in the week two this count is I mean it's now handling the 16 number of clients right? So similarly it has changed drastically on.16:58At least week three and week four, right? So these sudden changes right at the anomaly behavior it's being highlighted or it's being tracked by the AI AI cloud I mean AI cloud and it's going to give us sorry to disturb you. Your voice is a bit choppy.17:18Seems like a headset issue. Can you just try without it once? Still it's the same, I mean I just changed the mic now. Okay ok seems like maybe fan or something like that.17:36Maybe now I think it's better. Yeah, better. Thank you. Thank you, thank you. So what I was saying, I mean, these all our circles are basically the APs, right? And here what we are observing that there's a sudden change on the client's count itself, right? Here you can see in the week one, one of my.17:55Access point is handling the 32 number of clients. And suddenly if you see we have 16 number of clients exchange its behavior, so there's a drastically change on the number of clients count, right? So the capacity based it's measuring it out, there's certain changes. So these are very, very useful insights. I mean, maybe our one of the neighboring access points went down or.18:15It's it's changed, right? I mean maybe some behavior is getting changed, so those are also getting tracked by the AI cloud. So these all are the important insights which is coming by the AI under the trend deviations and it's going to do it either by the capacity as well as the throughput is getting changed, right? On the, on the wireless network, it's going to.18:35To capture those insights as well.18:55Next we have a network heat map. This is also a very important features, network heat map, so whenever we are troubleshooting a RF based problem, right? So this-that is great I mean, it's very helpful. We can measure it out the different, different KPI metrics like which of my access point have a huge interference on the network.18:59Per building wise, you can select it out any of the particular building with the location wise.19:19And you can select the band, right? Majorly the 5 ghz band which we are majorly focused, right? So, we can just check it out. What's the interference level on that particular locations? Right now I'm just selecting the global here. It's going to give you the insights like 6060 percentage of AP have high interference and on the left side you can see, right?19:24The how the interference is there. I mean the lighter means it's a less interference. Darker means the more, I mean huge interference on the AP and average it's giving it out that.19:4459 percentage is the average interference which is being found out on the AP and here you can see the per device, I mean how the interference was there on this particular AP itself. So whenever we are troubleshooting or we want to understand our network in a proper way, right? We can use a different, different KPI, I mean from the roaming event perspective, the fail onboarding events.19:53Perspective, right? We can select the different KPI and understand the our network and from the RF perspective, it's a very good information whenever we have to change the RF profile in a manual way and all those info, all those things. So these test lists are very, very useful for troubleshooting perspective.20:13Okay. And 3rd one is this peer comparison, so this is not a troubleshooting it's basically from the business perspective, right? How our network is doing, I mean we are comparing our network with the peer network comparison. So here you can see the, oh, I mean in my network perspective here we can see the blue colors.20:30This Highlights our network and the peer networks it's basically the similar network which other customers having and based on this PS data, just giving the reference data. I mean if we see right in my 5 ghz band, majority of the time my clients are around the 30 to 35 dBm and my PS network is doing around.20:5019 %, so we are much better our clients are having a good RSI signal, so we can just compare the peer to peer comparisons from the business perspective if they need this data.21:10And we have a network comparisons, so in the network comparisons, right? This we are comparing our own building from one building to the other building. So here you can see we are selecting the building one from London one and building two that's the San Jose one which we have selected and based on the different bands, we can select it out.21:22And there are also the different different KPI which you can compare the building to building comparison. So whichever is the I mean good performing building you can take those data and the worst performing building you can just do it out. So it's going to give us the informations about the different different KPIs, like if you want to check the interference, you can see which of the APs have.21:42The high interference, right? I mean it is going to compare the one building to the other buildings, right? So here you can see which of my 65 percentage of the APs have a low interference and 30 percentage of AP have a high interference based on this data we can go to the three 60 views and all and we can also see what is happening to our net.22:02Work or on those APs and we can do the troubleshooting perspective. But here we can compare the building to building and whatever the best policies are implemented on those buildings, we can just do it on the other buildings as well if those are performing very, really well.22:22So last one I just want to go to the baselining. So this is also very important. So here, what we are giving the data, we have a different different KPIs here you can see we can select more KPIs like DSCP time, authentications time. So whatever the circuits which you are seeing, right? That all are basically the buildings, so it.22:43It's not the APs or anything, it's all other buildings, so here we can see from the onboarding time perspective, my, this building is taking very less time. If you see my the two endpoints are there which is taking the 1.15 s onboarding time. And if you go to the last right on the right side here, you can see this is the my 1st building.23:03Which is taking onboarding time of 1.93 s, right? So it's just comparing it out giving you the data from good. I mean, I mean the worst side as well as the very good site as well. I mean the right side it's it's it's it's a worse site and the very good sites basically which is on the left hand side of it and whatever the circles which you.23:23You're seeing with the different different colors, right? Like orange or or or or the red colors, meaning there is some AI based alerts is being highlighted on those buildings. If you go inside it out, it will open that open baselining details and giving you the data.23:43So, here you can see, right? We have this onboarding failures and this green colors is basically I mean, baselining traffic of this of this network, right? I mean on this ID quest ID, these are the baselining traffic and the blue colors is the actual traffic that is flowing on our network.24:01So here from at this duration of time, our actual traffic, the blue colors, which is getting deviated from the baselining traffic, right? Meaning some is some anomaly is happening on this traffic based on this, I think there's an AI based alerts is being highlighted on it. So if you click on this, it will give you the more details, right? What's the problem was being encountered?24:21So suppose if I'm just taking it out this one, the onboarding failures on the side and here you can see the problem details, high experience, high percentage in onboarding failures is being found during this time durations and what it was the impact? I mean it's going to give you those informations what all the impacted.24:42Lines are there, right? These are the these all are the clients which are getting impacted on this durations and it's going to give you the root cause analysis, what's the probable net network cost, like authentications failure or the DSCP timeout. So those are the highlighted probable costs, I mean because of these reasons.25:01I mean our clients are facing the issues and I mean there's a deviations from the baseline it's giving us those informations and the actions also it's going to give it out for troubleshooting those network issues like we can verify the load of the AAA servers or we can verify the RF conditions or verify the CPU and memory load. So the.25:21These all are the very, very important insights which is which is given by the AI for troubleshooting any of the AI based problems on it.25:41So I, this is about the AI network analytics, which I have to cover, only just the last thing I would like to show you one thing, for this, just a minute. In the assurance, we also have issues and events page and.25:49There you can have a AI based alerts. Just a minute.26:06So if you see, right, at the in the issue sections page at the bottom, if you see, we have AI driven alerts and if you could just hover your mouse to the AI driven alerts, so all these are the AI driven alerts which is being highlighted in the cat C on based on the different, different locations. So you can go inside and do those analysis.26:24So on those wireless I mean wireless AI issues and all.26:44So that's it for the 1st demo. Now over to you Rahul. Alright, so now we have come to the use case two. The 1st use case was more on the wireless troubleshooting side. 2nd use case is on the RF optimization.26:49Okay and so in like modern hybrid working environments, the physical density and the client demands, it often shifts, right? And it's it's a constant thing and static RF settings cannot simply keep up. We need to have something dynamic when it comes to our RF optimization part.27:06So 2nd use case is on that, ok? Move to the next slide. So this is just giving an you know kind of a contrast or the comparison between the traditional RRM and the AI RM, the feature that we are talking about.27:26When it comes to traditional RM, it runs locally on the, you know, controller CPU or it is a reactive, often it makes certain channel and power changes during the middle of the business day when, you know, there is some kind of interference or there's some some kind of noise or interference spikes in your network.27:43It often operates like a black box. You don't have any visibility, it's limited visibility and NO troubleshooting requires a lot of complex CLI debugs. But when it comes to AI, it runs in the cloud.28:03So it analyzes two weeks of historical RF data. What it does is it anticipates usage trends and optimization channels and power with the full dashboard visibility. It also gives you a 30 day other timetable feature allows you to, you know.28:18See back, you know, past RF states and audit, you know, every single decision that you have made made, right? So not only it optimizes your RF, it also gives you a visibility part within your RM, right? What are, what are the kind of changes that you know the RFM has made?28:34And you know the in terms of dashboards, you know, all the, you know, RM changes that have made made in your network, it will be, you know, visually, you'll be able to visually see, you know, what kind of changes are being made. Alright, so this is.28:51Nothing but a, you know, short description or architecture of how AI algorithm actually works and the data flow. What happens is you know ok at 9800 controller, you know, collect the metrics from your wave to, you know, wave six Wi Fi six access points.29:10Right, what happens is the anonymized RF metadata is forwarded to the center and it gets, you know, sent to the Cisco AI cloud. The cloud engine evaluates the 14 day of historical data to, you know, uncover recurring interfaces or coverage patterns.29:27It generates an optimal RF plan and pushes the schedule down to your catalyst center. Now finally the controller updates access points you know during your configured offers, which which is a in a configurable parameter within the catalyst center.29:45And it delivers a stable, you know, flat free wireless experience. Now there are, you know, basic settings and advanced settings available, you know, when you configure the this all these controls are unified into a, you know, simple interface.30:01So under the basic settings, you select your frequency bands, whether it is a 2.45 or 6 ghz and you also define your site's busy hours. This tells AI when peak traffic occurs, so it gets telemetry during the day and apply changes during the quiet windows, you know, based on the.30:20You know, busy assets you have identified and configured on the center. And under under the advanced settings, you maintain the full control over maximum channel bits allowed the FSG channels and mandatory data rates. These are the main ones.30:38Now with that, you know, I'll again, you know, request to do a quick demo of them and then after that we'll be, you know, discussing about the AI endpoint analytics. Thanks. Let me take it up. Just a minute.30:56So now in this demo I'm going to cover the AIRRM piece and I mean, what are the things we should do it out to enable the AIRRM and what are the insights you will get it out after enabling the AIRM. So the very 1st thing it's very common that we have to enable, I mean in the settings page we should see that AI RM should.31:13Be enabled under the AI page, just a minute.31:32So under the Cisco AI analytics, we should see that our AIRM should be enabled under the settings page. That's the very 1st thing. 2nd thing, right? To enable the AIRRM to configure the RF profiles, AIRF profiles, you should go.31:36This design and there's a network settings page and under the wireless you will have a just a minute, you will see that we have a RF profile sections. So you can click on this RF profile sections and here you have the two two dashlets, one is this basic RF profile and second one is this AI RF profile. So if you.31:56You have provisioned the controller using the right so all those profiles is coming, I mean all those profiles you will see in the basic RF profile and I mean, if you have those settings I mean you have provisioned using the catsy you can just directly use this for converting this upgrading to the AI options. You can upgrade your AI profile. I mean if you have.32:16All the RF profiles deployed using a cat C And if it's not, so maybe you you are using only just for the assurance purpose, then you can create this AIRF profile. I mean from this you can click on this add options on the right hand side and this AIRF profile is there. So there you can give the profile.32:36And here you can see for which band you want to enable the AIRF profile, you can select those profiles, you can define the busy hours, I mean 04:51 P.M. where you have a lot of clients, I mean majority of the clients are coming. So you can just define the busy hours and we also have a busy hours sensities like low, medium, high.32:58So if you want don't want the much changes like channel change or power label during this period of time of the busy hours, you can select the low, I mean if it's a very critical environment, you can also select the medium if you want some of the changes you can I mean those channel changes or some changes you want it out on this period, you can do it out. Low means that.33:18It's it's not going to make changes on it only at a very critical scenarios it's it's going to make the changes during the low business. And high means I mean whenever it's it finds the improvement on your RF environment, it's going to make the changes on the power level or the channel changes during this time period as well. So it is just.33:38Very important informations, you can just use this informations bar and based on that you can select the busy sensity hours on your network perspective. And from the AIRM, what are the things you want to enable like flexible radio assignment, dynamic channel assignments or dynamic bandwidth selections, right? All those parameters.33:58This you can enable on, on the bands like 2.4 or the 5 ghz whatever you're using. You can select all those four or whatever you want it out to control using the AI, you can enable it out.34:18And if you go down, you will have a more advanced settings like what, what's the maximum DBS maximum width you want to configure up to the 80 mhz band you you I mean, you can do it out or or you can increase, but you can just use these advanced features, I mean advanced configurations for this AIRRM profile what.34:29So the channels you want to enable it out for this AIRM profile, right? And you can go, I mean what's the supported data rates for your clients? They should be doing mandatory, the minimum data rates you want it out on the client basis, you can select that and the TX power or there's a coverage hole detections. Those are the standard configurations.34:49Which you can use it out while configuring the AIRF profile. Once you configure it out, right, this AIRF profile.35:09Just a minute. I'm canceling because I already have it. So once you created this AIRF profile, right? There are two ways that you can deploy this AIRF profile. One, just you can go to the actions and you can select this design locations.35:16And there the workflow will get open and workflow is very seamless, it's very simple steps which you have to follow. You can go to the next and you can just select it out. I mean there are two ways. I mean if you are using only just for the assurance purpose, right? Your wireless and controller, you have not done the provisioning, so without provisioning.35:33Also you can deploy the AIRM profile and if you your wireless and controller is provisioned, then you can use this enabled with device provisioning, you can do that as well. So just it's it's just a workflow, you can just click on next next and you can deploy this AIRM profile, but the important caveat is there you can I mean one controller basis.35:53You will have a one AIRF profile. So it's not a mix and match, you can work it out. I mean, some sites you can work with the traditional RN profile and some sites you can work with the AIRF profile. It should not be in this way. So I mean you have to do it very I mean intelligent intelligently perspective that there should be the one controller which.36:13Is managing the sites we all the sites should be running the AIRM profile. So that's the important caveats. So there's another way that we can also use the workflow method. I mean, there's a direct workflow as well. If we go to this AI.36:33Just a minute, so here we will have a configure AI enhanced RNM proper workflow is there. We can either use this workflow. I think it will just open the page which we, which I just showed you a similar.36:50Really the workflow page will open and you have to just select the sites and select the controller and deploy the AIRRM configurations on it. So this also 2nd way it is also a very seamless way of deploying the AIRM.37:09So once the AIRF RRM is deployed, right, you have just deployed either using a workflow or or the page right in the AIRF profile where we can have those billing assignment and then you can deploy it. So once this AIRM is deployed, right? I want to show you how the statistics or what's the airm perfor.37:22Data we will see. Let me just go to the different lab, which we have at the D cloud one and I will just go to the assurance and here you can see in the AI network analytics page, we have the AI RRM and there you will have in, I mean, we can see the AI enabled sites, what are the sites which are.37:42Are getting enabled. So I have one European London site where we will have a summary how much.38:02How much APs are enabled on the AIRRM, what all the clients are using this AIRM profile? What's the RM performance, right? AP with the high co channel interference, it's giving us that data. So these one are the insights based on the RF perspective we will be getting in a one shot itself. I mean earlier in the traditional.38:10I mean traditionally we don't have any informations on the RF perspective, right? So here we will have a RF based information, whatever controller is doing basically on our network, right? So what are the changes is being happened? So the RF performance basis, if you want to see what are the what are times our RMS are getting changed, right? If you see the.38:30Yes it will give you the.38:50Full informations, I mean, what is being changed on this period of time. So if you see right there's a channel with change on this APs side. I mean two times this channel width was changed, and what was the trend of this changes, right? Here you can see majority of the time the channel change was happened. I mean at this.38:51Majority of the time you can have a full information about the RF informations at what time this channel with was changed, at what time this TX power was changed, right? Or the channel change was happened on this AP you will have those insights as well now on this AIRM perspective. We are also having a FRA changes. Similarly, we have different.39:11Different dashlets which is giving us the different informations about the RM performance, the co channel interference, right? And we can also see which of our channel is heavily utilized on our APs, right? So here you can have those informations, AP per radar, I mean and radar per channel, if there's any radar based frequencies and all, you can also have a look.39:31On those channels, if those frequencies are also there, you can have a look on this graph as well. And AP special density, I mean, what's the nearest, I mean, what's the neighbor access point distances there, right? So it is going to calculate those information as well. So majority of the time we we our APs are in the good, but we have.39:51Some APs which are very, very far, so you can have those special density as well and you can have more informations like if you need a service bundles and all so we can also use it out while troubleshooting if we need those data, we can download the service bundle and with the help of TAC we can troubleshoot as well.40:11And very important, right? I mean, after giving the informations about the RF information, it's also giving us the insights. That's a very valuable insight. If you click on this, it's going to give you how you can improve our network, right? It's going to give you the recommendations, so what the AI does it's given the observed like I mean.40:30We we have configured the 04:52 P.M. that's the observable but from the AI analysis it's 02:49 P.M. where the busy hours is there. It is going to give you those informations and you can change those busy hour settings using this page itself and here if you see the channel list, right, there are some channel lists, right? Which are.40:49Not enabled on our, on our controller, so we can also and I mean add more channels so that there will be the less interference on our network perspective, right? So we can change those changes as well. So after getting this data, we can apply it out, right? I mean DBS band we can change it out from 20 to the 40 mhz so all those.41:09Recommendations we can I mean we can if we are seeing this, these all are the valuable informations or if we want to change it out on our network, after analyzing it out, we can do those changes as well. Click on apply.41:29And you will have a, apply button as well and you can do those I mean you can apply those apply those AIRM profile settings as well. So this is the this is the demo which I would like to showcase on the AIRF profile. So now.41:42Over to you Rahul. All right, so the next topic of discussion is AI endpoint analytics. All right, so, so basically in point analytics is kind of profiling your endpoints. So basically it uses, you know, AI endpoint analytics aggregates data across.42:02These four primary sources, right? The 1st one is DPI props on your cat nine K switches. This is nothing but the packet inspection. It's a line rate the packet inspection using embedded and bar on your on your cat nine K switches.42:22Another source is Cisco ICE props, right? Cisco uses radius DHCP and HTTP user agents via PX grid to identify, you know, what kind to profile your endpoints. Also uses CMDB connectors. It says, you know, if you have integrated ServiceNow, it's a bi directional sync between Atlas Center and ServiceNow.42:39Find out or profile your endpoints. And the last one is the Cisco AI cloud.42:59Where we have machine learning models that group unclassified devices into behavioral clusters, you know, it reduces the unknown endpoints by 202 %, you know, if you have like any unknown endpoints and you are not able to profile them using any of those other sources, then using Cisco AI cloud you can you know.43:03Used.43:23Those insights from the Cisco A cloud to cluster those and profile those endpoints. All right, so when when you talk about additional profile profilers, right, it assign a single flat string like say e.g., this is.43:24Apple device, it'll only give a single tag to a particular device, right? But within point analytics it gives you four independent labels starting with the endpoint type. Example it could be city scanner or it could be a smart camera or a laptop.43:44And then we have, you know, the hardware model where it could be a MacBook Pro or it could be Optima, you know, CT 05:40, basically gives you a you know hardware model type. The 3rd one is the hardware manufacturer, whether it is, you know, Apple or assumptions.44:02And then is the operating system, whether it is running, Linux, right? Android, whatever it is, it also gives you that detail. Now this multi effector approach allows you to create highly granularize authorization rules.44:21And if you're using, you know, scalable group tags for the micro segmentation policies, it can also help you in that. Alright, so you know say you have a brand new zero day IoT device connect to your network.44:37At least its traffic characteristics such as open ports, destination domains and TLS ciphers and clusters similar devices together. This is the example that I was giving couple of slides back where if you are using Cisco AI cloud, you can profile your cluster those endpoints unknown.44:56Points together, right? You know, an administrator labels the group once and the catalyst center automatically generates the profiling rules for all the current and future devices which matches this profile.45:15You know, through our again our clouds cloud crowd sourcing model, when a customer A labels a new endpoint type, that anonymized signature is validated in the Cisco A cloud.45:34Now when there is another customer, say customer B which connects the same hardware, the catalyst center proactively suggests the verified label allow allowing the, you know, the entire Cisco global community to benefit collectively because, you know, you as an organization, you know.45:50Created those labels and any other organization, right? With the help of this community you know clouds sourcing with the help of the AI cloud can help everybody in terms of the profiling of the unknown employees.46:07Basically, this is the use case. Now with that, I'll again, you know, hand it over to Vijay for a quick demo of the endpoint analytics. Thank you Rahul, so let me.46:23Share my screen again. So now the last demo which we are going to do under the AI endpoint analytics. So let me go I'm expecting that we have already enabled the AI analytics under the settings page and for this AI point analytics we can see under the policy we have this AI point and.46:38Next page. So AI endpoint analytics gives the two things. I mean, it's giving the two purpose. One is that classifying the endpoints and second one we also have a trust course on those endpoints, right? So these are the two different things which the AI endpoint analytics does because it is giving us very useful informations, right? Because.46:58This majority of the attacks or anything which are happening to our network, right? That are because of unknown endpoints which are which we don't know about those endpoints, right? Those endpoints are getting targeted by the intruders for to attack in our network, right? So that's why it's very important to classify to what type of endpoints we have it in our network.47:18It is becoming more and more important in these days, right? So as Rahul already discussed, right, that we are categorizing the endpoints into the four major more four major informations. If we go to the endpoint inventory, I will just 1st of all show you this and under the endpoint inventory, right?47:38You will see.47:58Just a minute loading the data. So here we can see these all are the endpoints which are in our network and we have categorized each endpoints under this four category, what is the endpoint type? What is the OS type, hardware model and the hardware manufacturers. So if we have the.47:59Endpoints which have which have all these four informations, right? Means it is fully classified and I mean if if we have a partial informations means it's a partially classified and if we don't have the information means it's unclassified basically.48:19So let me just go you go to this overview page again and here you can see in the 1st page itself in the total endpoints, we can see we have the unknown, we have a partially profiled and we have a fully profiled endpoints, right? So, I mean these data it's going to give it out and how this catalyst centers.48:34Find this endpoints, it's based on the I mean three things if we go to the configurations, we can see those informations. So if you see right one is the C bar enablement, which is from this we are getting the informations, I mean those endpoints informations. Second one we are also getting the ice configurations from this profiles, right?48:54We are also collecting the data to classify the endpoints and we are also using ServiceNow integrations, right? If we are getting some informations to.49:14Profile the endpoints. So these all are the important configurations, I mean, which is required to classify the endpoints. I mean the 1st one is the C bar end bar. Second one is this eyes and the ServiceNow integration that also giving us the profiling informations on the client. So based on this data.49:24We are classifying the endpoints. Second one we also have a trust course. So the trust course basically we are also assigning each and every endpoints with the trust course, right? How much the trustworthy these endpoints are there? So how, how we are calculating this trust course I mean we have a lot of configurations, I mean.49:42If we go right, the very 1st thing I would like to show you this, just a minute.50:02So, so I, if I go to this trust analytics page, the 4th one, here the trust, I mean how the trust course is being assigned, it's based on the authentications type. I mean the endpoints which is connected to our network, what type of authentic authentication it is being used, like it's whether it's a doc.50:09Dot one X authentication, whether it's a map authentications, right? So.1X authentication means it's more secure, it's heavily secure, right? Mac authentications means it's less secure. So based on the authentication method, what it's being used, it's going to check that. Second thing, whether the posturing on the endpoint is being done or not. If the posturing is being done means it's.50:29It's more secure based I mean if it's and if a posting is not done on those endpoints, it means the less secures. That's the very 1st authentications method as well as the posturing thing, which is being checked it out to give the trust course on it. And second, we are also checking, I mean.50:49I mean if any spoofing attack is happening, right? I mean, any max spoofing or those those kind of vulnerabilities being found on those behavior of those endpoints, right? So based on that, it's going to decrease the trust course on it and if these are the things which we are going to do it out if we see the trust score.51:08Right I mean if there's any what are the things or capability we can check it out on the I mean on the trust course I mean we can also see the AP API spoofing detections if there's any max spoofing, it can able to detect open port and credentials vulnerabilities is normal, I mean it's already been disabled from the past versions, so.51:28Only if we have the AI spoofing detections, we have a network context, right? I mean NAT more detections concrete Mac address and if if any of the endpoints is trying to access any IPs, right? I mean which are unsecure IPs on the network. So based on these, if any of the endpoints are trying to, trying to attack it out.51:47On our network, the AI endpoint is able to detect it out and based on the trust course it's going to decrease the trust course of endpoints as well, so.52:07So these are the things, I mean, I mean once the trust was is getting decreased, right? Then we can push the policies as well based on the ANZ policies which is created on the ice. So in that way we are securing our network. So that's how, whole AI endpoint analytics features works, I mean with the along with the integrations with the ice site.52:18So these are the things which we are getting on the AI endpoint I can't complete all the things right now on this, on this demo itself, but I'm just giving you the highlights of it. Like it's, it's going to do the profiling and second thing it's also going to give the trust course on the endpoints perspective. So these are the things on the AI endpoint and which I would like to.52:37Cover in this demo. So thank you. Now over to you last year.52:58No worries. And now time that it's time for like I see that we have answered all questions.53:04Team, keep the questions coming. If you have any questions, please post it on the chat or on the slider and just a reminder, so once this once you end the session, you will be redirected to a feedback survey. If you could.53:19Please let us know your feedback, that would be really helpful. We will be there on the call till eleven.53:38Yeah, sure. Go ahead. Let me share the so for all the recordings, you can find this link, I'll share that.53:56So once this session is ended, you will be receiving a mail and this recording also will be uploaded on this community page. And we do have all the past recordings in this space.54:13