Deep-Dive into CNC-EMF

Okay, thank you everybody for joining.00:10Can start now. So yeah, welcome to session two of our expert inside series demystifying CNC EMS basically yesterday we established some foundations of the element management function and how it provides.00:16Basic infrastructure visibility and how the, what we saw is the other management capabilities also. Today we will be going much deeper into it. So we are going to explore some inner workings of CNC EMF focusing on how it manages deep inventory.00:36Discovery, how it handles granular performance monitoring, how it implements service health assurance based on modern RFC standards and how it automates basic complex software image management workflows.00:56So, thanks for joining alongside my colleague Dilip and Patrick are there. They're going, they will be co hosting us today and we are going to guide you through this advanced topics to help you transition from.01:13Some of the basic network monitoring to proactive intent driven operations so for those who may not have joined us yesterday.01:31Let me introduce ourself. My name is I'm a customer success specialist here in Cisco and means joining me is my colleague, who is the customer success specialist with CX along with Patrick.01:48So our primary mission in CX is to partner with our engineering and operations team to basically ensure you are successfully adopting or scaling to Cisco automation platforms. So we'll be guiding you through both the technical presentation and demonstration.02:06Our video which will guide you through guide you through video demonstration and he will also assist you in between. So let's start today with some logistical items I will just go through.02:25So if you have any questions, there's a Q and A panel with a.three.menu along your right hand side of your scheme, you can use that hand icon to raise your hand and we'll try to respond to any of your queries.02:41So let's review our agenda today. So we have structured this deep dive into four core chapters and one of it is more 5th one but it is a more detailed version.02:59Will be covered by the 1st we will explore inventory management looking at how EMF performs deep discovery or deep inventory discovery under the hood. Secondly, we will cover performance monitoring, detailing how devices collect counters and config.03:14The threshold crossing alerts. Thirdly we'll discuss about discuss about service health, so explaining how CNC aligns with a well known RFC 9417 to provide an end to end service visibility. And 4th, we will examine the swims suite.03:34Detailing in parallel upgrades or platform support how to build your software library, and then finally we'll wrap up with custom metric solution and a demonstration with question and answer sessions later on.03:53Let's begin with the 1st topic inventory management. We will look at how CNC EMF moves beyond basic device listing to provide you with a deep hierarchical discovery of your physical and logical network.04:12Assets. So at the individual device level CNCMF provides a dedicated inventory tab within the device details view as you can see on the screen. This is not just a static list of parts, it is dynamic. So hierarchical representation of a device.04:27With its physical and logical components. The inventory automatically updates whenever there is any physical or any logical change detected on the device. So if you can see it means you can think of an example that if an operator hot swaps.04:47A faulty optical trans receiver in a core router. EMF basically detects this as a change via an SNMP or a netconf, queries the new trans receiver's serial number and model and updates the UI automatically.05:03So you can use this built in filter options also to quickly search for a specific component such as searching for all your optics that is hundred G or ten G, maybe high density line card can use that as well.05:18So when we scale this up to a network level, CNC EMF provides at a glance view of your entire total network inventory. This dashboard allows you to immediately identify where your hardware assets are.05:36Physically located and across your olderpology. So if you receive a security advisory example for the hardware recall notice or some specific batch of line cards, you don't have to log in into dozen of routers to search through manual spreadsheets or something like that. You can use this network inventory dashboard to search for your affected.05:52Part number or the serial number and instantly identify what exactly the router it is, where it is housing, where those cards is. So you can use this as a great network asset or a network inventory asset.06:12So this slide shows you a visual representation of a detailed inventory interface. So this is a very, a beautiful feature in in CNC EMS, so as you can see, the UI organizes the inventory into very clean and nested tree structure.06:30You can expand this to the chasing level to see what all individual slots you have. You can also expand the slots to see what are the line cards and drill down further to see what physical ports and trans receivers you have into your network or saying your device specifically.06:48The filter options at the top of the screen will allow you to slice and dice your data by operation, by operational status or admin status or hardware type giving your operations and procurement team a very highly accurate real time audit of your network assets.07:05So now to reiterate what we went on through the last slide, the key value of EMF device inventory is its dynamic nature. It's, it's tightly integrated with our change detection mechanism. So that's the main.07:28If I can say key value which we provide, Whenever a physical card is inserted, the port is reconfigured or sub interface is created and the inventory updates itself automatically. So this ensures that whenever your operations team is troubleshooting on an interface issue, they are looking at the exact.07:45Current physical status or the logical state of the device, preventing any confusion into, the troubleshooting or in the configuration maybe. So now let's look under the hood how this EMF deep inventory discovery process actually works.08:03Okay, so it's highly little bit technical process that runs on the background. 1st the discovery process is not exposed in the GUI. It runs in parallel with the CNC device lifecycle management process called the DLM.08:22But because it collects highly granular data, it can take longer to complete. So you can monitor its progress and retrieve results via our REST APIs. If if any of the node discovery results in kind of say collection failure state or NO data is written into a inventory tab.08:37If it results in kind of a completed with warning, you may still see the partial output for it and depending on which inventory feature is filled, you can just look into the inventory collection status. So architecturally EMF discovers or discovery basically leverages standard MIPS.08:56Specifically SNMP V two MIP entity mip, and if mip. That is interface.09:15It executes on demand dynamic jobs at the crosswork data gateway and these jobs run via modular XDE packages leveraging the data gateways SNMP collector. So which which ensures that all the heavy lifting of the data collection is basically distributed and.09:21Doesn't impact on the.09:41Performance of the CNC controller. So now let's transition over to our second topic that is performance monitoring. We will discuss how CNC EMF collects and analyzes the device level counters to help.09:42To maintain optimal network performance.10:02The primary goal of CNC performance monitoring or the PM is to focus on device level contacts. This distinct from this is distinct from service level monitoring which we have, which we will cover in the next topic that is the service health section. So our performance monitoring policies allow you to monitor a.10:09Wide range of critical technologies and hardware areas.10:29So if you can see out here, in the slide we have device memory and CPU counters. This, this basically ensures that the control plane is healthy. Then we monitor internet interface health, which tracks the input output errors drops, and utilization.10:33We have optical counters monitoring optical transmission and receiving receiving power levels, then optical ZRP counters for high speed coherent optics and PTP and sync e counters also basically for critical timing and synchronization in mobile black backhaul networks.10:51Traditionally GNSS counter also to monitor satellite receivers signal strength on your timing notes. All of these.11:11All of these policies, right? Are feature extensive and customization options are available, allowing you to basically further tune the collection parameters to ensure or say match your specific engineering standards. Performance monitoring.11:18Basically, it isn't just about real time telemetry, it's about an historical context as well.11:38So when we see as a performance, we have to look at as an history say one month, one week or say one year basis also. So EMF maintains a robust performance monitoring history that is accessible across three views. The 1st one is the device details.11:43So the 1st one which we see over here is device details, link details and interface level details. So you can use this kind of which isn't associated with an example, say if a customer reports that any experienced.12:01The experience of intermittent packet loss between say midnight between two and three M you can open basically a link details and or the interface details view and look at the historical performance charge for the specific window and you can correlate the interface utilization whether is there any any count, any error counters.12:16Which is visible or which is visible or at optical levels you can see the physical layer issue, is it a traffic spike or what's the root cause basically for this issue?12:36So you can leverage for those kind of troubleshooting. So to start collecting this rich performance data, performance monitoring collection must be enabled on a per type basis within the CNC administrative settings.12:46So we go to this yeah and for every collection we need to create some policies out there. So one important technical note to our customers and operators is that you will see a policy option labeled as LSP traffic in the interface.13:03Please note that this specific policy is reserved for Cisco CX internal use and troubleshooting and it is not supported for end user configuration. For all your operational needs, you should focus basically on enabling the device interface and optical level and timing policies.13:19To basically further monitor the performance parameters. To make performance data actionable CNC EMS provides with a granular control over polling frequencies and thresholding crossing alerts.13:37Called the TCA. So you can define the KPI granularity and polling frequency based on criticality of the node. E.g., if you might pull a core router interface every 30 s while pulling in edge switch memory every 5 min.13:56So it is not much of a load to the device or the CNC. So furthermore, you can configure the TCS with multiple thresholds such as warning major or critical and applied basically certain to it.14:14So if you set, say example, if you set a CPU utilization alert at 80 %, say with 5 % distances, the system will trigger an alarm when the CPU hits 80 %. However, it won't clear that until the CPU drops below 75 %.14:31So this would prevent an alarm flapping issue. So where fluctuating matrix constantly triggers and clears your alerts driving your operations team crazy basically. So yeah. So this is a very good feature for TCS.14:50And for those who are planning to upgrade CNC to seven.one, there are some important changes to this default performance monitoring collection behavior that you would need to be aware of. In release seven.one, the LSP traffic and the interface health policies are now enabled by default.15:09If you are upgrading to a version seven.zero to seven.one of the CNC and you did not have these policies configured previously, the upgrade process will automatically create and enable them by default. So yeah, please note that modifying or disabling this default policies.15:29Can have impact on your crosswalk optimization engine or CNC capabilities. It can also affect how visualizations and the data are being displayed and rendered within your topology user interface. So it's, it's highly recommended that keeping this default policy active is highly recommend.15:49So now coming to our 3rd and the most I can say, agreed USP for the Cisco's CNC.16:14It is the service health. So we will examine how CNC EMF implements the modern standard based service assurance, and creates a bridge and bridges the gap between physical network health and the actual customer experience. So this is a very good interesting feature out here.16:26So, service health is built on foundations of RFC 917 which defines usually service assurance for intent based networking. So if you remember in the last session, we spoke about autonomous or autonomo autonomous network, the autonomous network.16:49It is required to have in the 4th and the 5th stage of any autonomous network to be intent based network. So it is an initial step towards that step four and step five. So the challenge usually what we consider is traditionally provisioning a service and monitoring a service.17:09This were kind of decoupled. So when you provision an L three VPN and you monitor an individual interface, this would create a disconnect and you don't actually know how the customer service is performing. So you you you are able to see that, ok, L three VPN is up, my interfaces are up, everything is up, but customer service.17:29Customers experiencing problems is having some kind of experience issues. So this was leading to a high time to identify that what the issue to the customer is. So that's what the mean time to identify MTDI.17:49And the meantime to know. So what the CNC came up with the solution is that it takes a service centric approach over here because we have a full view of the network. We have an inventory, we have a very detailed inventory, we have KPIs, we have everything, right? All informa.18:05Services information CNC is capable of driving all the services, creating service man, using those services, right? So we have all the information. In this case, the service health takes a service centric approach over here.18:26It dynamic dynamically ties all the service in that what you provisioned to the underlying infrastructure telemetry and it codifies an export troubleshooting know how with the automated rules. So the outcome is that.18:39It reduces time to detect and remediate the self level service level issues, means ensuring a very high superior user experience. So let's look at how this transforms the operational experience for your.18:54Support teams. So example a customer calls your frontline support complaining about a connectivity problems or so. So in a, in a traditional model, level one operator would have to escalate it to the level two or level three engineers to run basically trace routes and check the CLI outputs.19:14But with CNC service health, if you have the CNC service health, the level operator has an end to end view of the service. So they can immediately see what device supporting the customer area is dropping packets and the system will automatically troubleshoot the issue using the heuristic package. So we will discuss what is the.19:33Juristic package in the next slide.19:53So it protect proactively notifies level two and level three team with exact the with exact root cause so keeps your highly skilled engineers basically focus on only the remediation part rather than basic troubleshooting part. So.19:56So understand the service health works, we need to define our four key terms that is as I told you, right, heuristic package. So what is that heuristic package? So this is a set of all predefined rules within CNC that automatically monitors the services and troubleshoots common failure notes.20:13Okay, and and the assurance graph is the dynamic hierarchical dependency graph of everything that feeds into the service health. This, all these components are called sub services. So an assurance graph is a basically a display of all the connected services.20:34With its own nested services which are working to get you the service running with level one, level two kind of all those details e.g. if you're saying, say my ping level or the interface level and then at the routing level at the session level.20:52Will then at the application level all that whole assurance at the service level, how basically the the assurance graph will display will see that in in the.21:11In the demo as as well, you will have a look at it. So 3rd is the monitoring levels. Services can be monitored at basic or the advanced levels to advanced levels collect more KPIs, but requires more.21:23System resources. So that's the thing and the 4th is the soft dependency. So these are the sub services that are of interest to an operator like the router CPU or memory, but won't directly bring down the service state. So that is also taken into consideration. So it's a soft dependency if I can.21:38Dotted lines, right? We come, we usually say a dotted line. It's a dotted line as a soft dependency to the its own router CPU and memory also.21:58If they are unhealthy, we are selected also as a soft dependency, so they don't trigger or raise false and service downloads.22:07Sorry, so this slide illustrates the power of the assurance graph, which we discussed in the last time, right? Definition of assurance graph. So this is how it will look like in the CNC UI. When a service health basically turns yellow or red, an operator can click on the service.22:24And drill down onto the graph. So this is how the drilling down would look like. So that is a one service over that service it is say layer three, how the basically all the routers are connected, then all the interface health, all the packet level details, everything.22:42E.g., the graph visually maps out all the sub services also such as the physical interface, routing protocols, tunnel paths, it's a BGP neighbor session, any kind of flaps the specific sub sub service node in the graph will turn red. So.22:59So if you can see some, that over here in the graph, so you you will be able to basically detect that because of this reason, the session is flipping and the sub services note is down and the graph will turn red.23:15So it allows the operator to instantly see what exactly the service is affecting. So another another good thing is that service health also includes historical tracking.23:31So by analyzing the past history of the service health state, you can easily identify recurring patterns. So if example say service degrades to a warning state at exactly 10:00 P.M. every night and recovers at 10:30 p.m.. So you have this.23:47Issue ongoing, so you can look at the historical timeline over here to correlate its behavior with the scheduled network task which you have and such as the automated backup jobs or night batch transfers are allowing you to resolve any chronic or intermittent issues.24:05Oh next next topic is our final, not final, it's a one before the final basically and it covers the software image management and we will look.24:21At how CNC automates and scales the software upgrade life cycle across your network.24:41So partially we had covered this topic in the past session. So to recap what we introduced yesterday CNC swim is a, is a built for basically mass scale operations.24:48So right now it supports up to 25 parallel upgrades and utilizing secure protocols like SFTP SCP as well as FTP and DFTP. It supports in service software upgrades and to eliminate any kind of downtime of or compatible performance platforms.25:03So crucially swim supports both standards, that is the ISO images and the golden ISOs specifically for your.25:24Lint architectures which include NCS 54540L ARM based and Cisco 8000 NCS 5700. So this ensures that you can maintain standardized pre validated software baselines across here.25:32Most of the generation of the access and the core networks So these are the platforms which are supported. We discussed this in the last earlier session as well.25:48So, so the the list of the devices gets increased and the parallel operations also gets increased once the, once we move to a different levels of CNC, so that is seven.one, maybe 25 seven.two is having 50, so you can just check on.26:03Into the versions compatibility and then decide on.26:23What needs to be or how much needs to be parallelly upgraded. So this was also actually done in the last slide on the last process is driven by an intuitive four step wizard process.26:28So we have to select image and deployment option is defined, we have to select the devices, validate the image, and then this is a kind of a structured workflow which ensures every upgrade is executed very safely and consistently. So once a job.26:45Operators can monitor its progress in real time over here. So once the job monitoring dashboard provides a complete job details showing you exact status of each phase of your upgrade, so such as file transfer, installation.27:06Device reboot or for every target device if an upgrade fails but if and by any chance on a specific node, the system will hold the process and display details as error logs allowing you to quickly troubleshoot the.27:22Affected, upgrade process. So to perform the upgrades, you 1st need to build on your software image library.27:38So CNC provides with multiple flexible options for uploading the images into the database and 1st you can upload the image and then directly from your local client machine via the GUI. Then secondly, you can pull images from the remote server, you can use SCP SFTP or HTTP.27:53This and maybe you can also use legacy devices also, you can import the image directly from an onboarded device if you want. So this is already running the target software version, so this flexibility makes it incredibly easy to populate a repository and maintain a single source of truth for your.28:13Network software.28:34So, yeah, custom metric collection, this will be covered by Dilip and further he will be going through the demo section with you. So over to you then and thanks everyone for joining.28:37So we will scroll the slide for me, ok? Yeah, so I mean explained about that there are many parameters and many things which which can we can monitor.28:55The which is inbuilt or it is inside default inside the CNC itself, whatever that general things or what all we thinks that is what it is available. But there are some circumstances arises, let's say you are upgrading your some device or some services you are changing.29:11And if you want to monitor that specific service, then someone need to sit and then he need to monitor that services using some CLI and some commands and other thing and then need to pass the data and they need to understand that what exactly things are happening.29:30So that kind of custom thing that that is what I mean day to day life or some issues and some your your director or someone come and they'll ask, why cannot you monitor this thing? Why this parameter is not available?29:46So CNC gives that flexibility to you that you can define your own custom metric parameter. You upload it inside the CNC define your parameters and what all data you wanted to pass, whether that is what you can do.30:03And then you can generate a report. So what happened is that as I told you one is that the service monitoring or another thing is that upgrade and post upgrade you want to monitor something. So that that's where CNC is very much useful for your day day to day job using this custom data collections.30:20Okay. So it I mean this one allows you to do that static data on the static collection and which is not collected by default from the any device or because whatever the default that that is what it will be collected as defined inside the CNC itself.30:40And then custom is not stored. So you you cannot even do that pulling using some API because if it is stored, then you can just go ahead and pull using some APIs and then then you can leverage. But this one is that specifically thing particular to device or service.31:00That is what you can define the parameter, and then you can start pulling and storing inside the CNC, ok? And how it can be done so you have a lot of flexibility that the collection can be done.31:19Sorry, yeah. Collection can be done using SNMP and you need to define your Yang model or you can use the GNMI which is GRPC or also you can go with the telnet SSH or netconf. Okay.31:34So I mean it it and again it's not limited to Cisco only because when you are writing the custom things and when you are putting via SNMP, so all kind of devices, whether it's a Cisco or non Cisco, that th that is what it is support supported using this custom code because.31:50When definition needs to be done, that that's where you define that, ok, this is a Nokia device or this is uniper device and this is how the parameter needs to be collected. So that that is what you can define and the the criteria, the device criteria that is what you need to define when you are the.32:10Creating the this rule or this policy.32:30And the data will be kept inside the database or CNC because we all know that CNC itself also keep collecting the data for hundreds of thousands of devices. So for this custom thing, you need to take a note that it is by default 24 h but when you are uploading.32:33It gives another level of.32:53Granularity that you want to maintain is for 24 h or 48 h, so which is also trickable parameter that you can go and define that, ok, I want to do it for three days. So just go when you are uploading inside the GUI and then you can go and change that what all you wanted to how long you wanted to retain the data.32:55Okay and data can be plotted as a graph as well as if you want to do that inside that CSV format or histogram or top end, so depending on that what where you wanted to plot it inside because CNC has the interface as well as.33:16As you want to do and fetch and send this to as a.33:35A picture or a CSV file in the tabular format, so that that is what you can do and then you can define that what are the top talkers or histograms or summaries used. So depending on the definition that that is what you can do. As well as there is a one more options is available that from CNC you can.33:39Create your young model and then you can do that rest API also. But that's where you need to define that what all threshold and other thing and normally when I'm referring CX, so ideally, we recommend that I mean if you are not comfortable, then the CX team.33:59The which is the Cisco professional services, they will help you out to do this kind of thing. So customization either you can do it by yourself, but if you want to bring the expertise in, then CX can define and help you out on the this thing.34:19Okay, so CNC basically gives that flexibility to do that customization, NO need to run that scripting and other thing but you do your scripting and CNC will pass the data collect the information maybe you can move to the next slide.34:35That how it can be done. Yeah. Yeah. So now the question will come that what all needs to be done. So the 1st step is that what needs to be collected what is your objective or what all things you wanted to be, you are going to collect. So that's what you need to define that this is my object.34:50Let's say I want to do the power collection or I want to monitor the sensor data from the fan you have let's say five fan and one of the fan is giving some problem inside your chassis.35:10But you want to define that traffic parameter? So that is what you need to decide that what all parameters you you wanted to monitor. And when you are defining, then you need to choose that which protocol is best suited for this kind of collection.35:21So that that's where the expertise is required that ok if you cannot define by yourself then you you can get to engage with the CX team and they will be able to tell you. So like let's say if you want to collect that as I was mentioning that one of the is not functioning properly out of five inside the.35:38Then for sensor data, basically you need to go with GNMI and let's say you are doing that SNMP so this is a very known thing that oids is the thing that that is what you can leverage that object ID and that's what for all vendor OID is available and.35:59That is what you can leverage to collect that information. But let's say you want to do that command level things, then obviously in place of human being monitoring or collecting the data, you can go with the netconf or telnet or SSH base and then you can define and enter.36:19Into the devices and then collect the the whatever the information you are looking for, the custom parameter you are looking for, ok? Then you need to define that how you are going to collect that the data. So.36:38For each metric, you need to define that KPI attribute that, ok, let's say my I I am monitoring the power supply and when power supply is hitting let's say 72 degree and I want to monitor that thing. So that that's how you need to define that KPI metric.36:53That what attribute or what all attributes you are going to, let's say you are monitoring the traffic and you want to see that when it's picking up or when it is going down or CPU when it's suitting some threshold. So that that is what the KPI attributes, that is what you need to decide.37:11Decide. And then store race is obviously an important thing that that's what needed to be considered that how much because if you are collecting so frequent data, then storage will also run out of the capacity because storage is always limited or depending on.37:28How how how the deployment is done. So that's what you need to make that decision that how much data, how frequent, and that's where the KPIs or KPI attributes plays the key role, ok? And then define that once you have decided that what all record you are going to.37:47Collect and then data will be translated inside and it will be collected as a model dB because whether you are going to collect when you are collecting let's say CSV file, then you need to define that, ok, between this time to time period or this is the data, the peak.38:07Low high and whatever that KPI parameter, that is what the columns names that that's what you need to define and then description and then which all devices because in your environment you are running thousands of devices, so you don't want to collect from everywhere or.38:26Everything and then which vendor or manufacturer so that that is what the important parameter needs to be defined. Yeah, so we can we can go to next slide.38:46Now you have decided you want to monitor something. Next step, you define that ok this is what I'm going to monitor. The important thing, the next question comes in that I can just go and write and deploy this one.39:06Yes, you can but for doing anything, you require some basic rules, so for the development environment, you need to have that git repository. So you need to have that 3rd party Git repository that that is what it is required.39:20And Java and install and Java for for that java runtime. So that that is what it is required. Okay, and CNC device, I mean it's it's a non brainer but CNC setup with required devices.39:37And will be used to debug and test the package because you will not not just go and deploy directly inside the production environment. So it is highly recommended that validate your script in place of bringing down something. So if possible inside the lab environment you can do, but in the product once the.39:54You are preparing your setup.40:14That's where this and get repository that that is what it is required. Because in the future also you want to reuse your code, then Git repository will help you that ok on this on top of this one or let's say you have created one version and then on top of that when you want to enhance something or on your.40:16So that that's where this will be used.40:36Okay, now starting to write your model. So the 1st thing you need to is do is that specify the schema names, ok? And that is what your anchoring point or when you are getting inside your GUI.40:40You will be able to see this, this is the my schema name. Let's say I'm calling that disk monitoring. So I will call that disk bone or whatever the user friendly name which you can understand that that is what you need to define as a schema name, user understandable and that name.40:56Will appear inside GUI.41:16If you are collecting the data using columns or size, so that that's where you need to define that, ok, my data will come in this big or five digits, four digits, whatever the digits are in bandwidth or whatever the parameter, so that's where you need to define the.41:19Column names, then what type, what size in case of it there, ok? So if it is a string type, then what size the strings you are going to collect so that that is what you need to define inside this model.41:37And the important thing is that which dashboard because yesterday if you remember Hussan was trying to showcase that you have custom fixed flexibility to create your own dashboard. So would you like to put on everyone's dashboard or you want to do it on your own dashboard, you are creating some custom dashboard.41:54Something for demo.42:13To look into that one. So that is where you need to define this that where you are going to plot it or chart it and then when I say TCA that that's where the threshold comes and that is what you need to define. But if you are not sure, always go and talk to your customer experience or problem.42:14Personal service team and they will be able to help you out to specify the value or range or unit types.42:34So that that's where they can help you out or you can engage them for this kind of things. Yeah we can go to the next one. Yeah, so.42:42This is where I mean I have just tried to showcase one simple example that I would like to monitor the dispace and as I told you, I defined the 1st thing with the disk space and this UI will be displayed as the display name when you are getting inside your.43:01The the CNCGUI. And then the key attribute I I need to define that, ok, whether I want to location so basically location is that where which RSP or which switch where where exactly so that that's where I need to define.43:17And then let's say here I'm clicking but I think it's not. So I want to monitor let's say free disk space, so I need to go and collect the disk space size disk space utilization. So such kind of thing that that is what it is called as a parameter.43:34And that's where it will go and fetch this information and then it will calculate and it will plot inside your dashboard and then either you can define in the exact value or you want to define either percentage.43:51Then you want to define that, ok, what is the minimum or maximum and that that is what you can do and that health condition you can define that when it's reaching let's say at the right hand side bottom, you can see that you want to define that healthy criteria.44:11So this is how that young model can be written, that minimum is 0 % maximum is 50 % shift let's say it is going above 50 %, then you want to see some kind of alert. So depend on your view, the custom view that that is what you can define.44:28Yeah, maybe we can go to the next one. Yeah, and then implementation. So which device because we don't want to I mean depends on that how many devices you want to.44:48Collector across the network. So that's where you can define that which device, whether it's iOS xr. So here I'm showing the iOS XR and then NPU. So that is a network processor engine because that that's what it works to collect that information.45:02Okay because on this RPs it is called NPU network processing unit. No, not the CPU, but yeah, so that's where I have I need to define that NPU and here you can see that this space on the 3rd yellow color.45:19And then the the KPI and then run generic CLI command, then procedure is CLI, you can see that it's a C CLI. So that that's how it can be done. It is this one is a very simple example, so.45:36You can easily understand and as I told you various protocols are supported so you can see that protocol in the previous slide we have written that young model, so I'm calling out that this is a young model. That's what I'm going to use. And when doing this young model I need to define that MIB.45:53And this is what the Mib type so yeah it depends on that what need to be monitor. So that is what it can be defined and then it will start parsing the data and then it will showcase inside the.46:12Yeah, so with that one we will go to demo. I can go to demo, right? Yeah yeah yes, ok.46:32So we have, we have got three demo for you. One is the device and inventory management, that is what we discussed today. Second one we discuss about the performance management and the 3rd one we discuss about the health.46:47Service health monitoring. I'm sorry with that one, let's go with the start with the inventory management. Welcome to device and inventory management in the Cisco Crosswalk Network controller. The dashboard summarizes device status, service health, and traffic engineering. Twelve device.47:04Devices are managed with reachability and operational state. Let's open the topology view to explore the routers.47:24The map plots each router with the inventory listed alongside. Eleven of twelve are reachable, each with address, host name and time. Select. You are able to hear the voice, right? Yes yes, Yes, ok.47:31Ting Node two opens its device details. The summary shows device type, product family, series, and software. The device runs iOS XR with a direct SSH session.47:50Scrolling reveals its topology and routing attributes. These include location, device group, and routing identifiers. The ISIS system ID and AS number placed in the routing domain. The linked tab lists every link on the device, seven in total.48:04Each shows linked type and state with a side and Z side interfaces. The interfaces tab lists all 15 interfaces. Each shows administrative and operational state, streamed live. Down interfaces are clearly distinguished from those that are up.48:23Inventory stays current through telemetry, not periodic scans. The history tab trends device health and CPU. Detailed inventory reveals the physical hardware hierarchy. The chassis holds a root processor and a line card.48:43Yeah, this is this is what Hussan was trying to showcase that once you are clicking on that node level, you will be able to see that it is running that iOS XR and then it's getting inside that low level information. So it it will it will go up to that interface level and deep dive and it will be able it will showcase that deep level inventory.49:05Each is modeled as a component discovered automatically, and wherever you are seeing this plus this model supportsolation. RMA and capacity tracking.49:25Expanding the line card exposes its gigabit ethernet ports. Each port is an inventory object linked to its interface. This joins physical inventory with logical state in one view, selecting a port shows its traffic history and health.49:36Its properties list state, description, MTU, and speed. Here the port has an MTU of 1514 bytes. Returning to topology, the map reloads the current state.49:55All routers in their inventory appear again. The topology reflects the current inventory and state. Selecting a link opens its details. This link joins node five and node four and is up.50:12It is an ISIS, IP version six, layer two adjacency, the summary shows utilization, packet drops, errors, and metrics, IGP delay, and TE metrics steer segment routing across the link. The history tab then sharps the links traffic over time.50:31In summary, Crosswalk gives a live view of inventory and health. It is inaccurate, always current source of truth. Thank you for watching. Yeah, so next one we will go with the performance management.50:50Welcome to performance management in the Cisco Crosswalk Network controller. The topology view maps the live network driven by streaming telemetry. The inventory lists every router with reachability, address, and product type. Eleven of twelve routers are reachable. One is unreachable.51:13Device state is tracked continuously without manual polling. Selecting a node opens its device details and summary. The summary shows device type, product family, and software version, a direct SSH link connects to the device in context.51:32Tabs cover links, interfaces, alarms, inventory, and health history, data is streamed by model driven telemetry, not SNMP polling. The history tab trends device health over time.51:52CPU utilization averages about 28 %. Memory pool utilization averages around 45 % counters stream periodically, showing resource pressure, the window can narrow to the last hour for recent behavior.52:07Yeah, so I mean this one also gives that another level of flexibility that you can just go and define that what all parameters, so was describing that device health and lsp traffic.52:32You can interface help.52:45So that that is what you can go inside this device management and then you can define your own policies of parameters and accordingly CNC will monitor your performance data. Active policies gather device.52:46Space metrics. Each policy binds sensor paths to a collection cadence. Returning to topology, the map reloads the current state, all routers and their inventory appear again.53:06Selecting a link opens its details. This link joins node five and node eight over gigabit ethernet. It is up using an ISIS, IP version six adjacency. ISIS also distributes the prefix SIDs for segment routing.53:17Details show interfaces, router IDs, addresses, and utilization. Utilization comes from streamed counters against link capacity. Both directions are currently likely used, the history tab charts outbound traffic rate over time.53:37Traffic holds near 670 kbps. Traffic delay and loss are trended to reveal congestion. Packet drops remain at zero, a healthy forwarding path. The full history supports capacity planning.53:54Yeah, so here I mean once you are here inside this one you can you can download this report as a PNG or you can also download as a CSV format. So if you want to do the tabular way, so this gives that flexibility for you to download the PNG or CSV and then you can report to your customer or wherever you would like.54:11Sockets to executive levels. Charts export has CSV or PNG for a flying analysis. Files are written to the local downloads folder. The CSV holds the raw time series per direction. This feeds external reporting and capacity.54:32Jaws.54:52In summary, crosswalk turns telemetry into performance insight. It enables proactive operations and capacity planning. Thank you for watching. Yeah, so next one we will move to this service health, which is very important when we.54:54Welcome to this Walkthrough of Service Health assurance in the Cisco Crosswalk Network controller.55:14Crosswalk delivers closed loop assurance for layer three VPN services running over a segment routing core. The home dashboard consolidates the operational state of the entire network into a single view. At the top, we see VPN service counts alongside the health of each monitored service.55:20One layer three VPN service is being monitored and it is currently reporting a degraded state. The dashboard also surfaces transport slicing and traffic engineering, including segment routing policies here Eight segment routing policies provide the engineered transport paths between the provider edge routers. Let's.55:38Navigate into services and traffic engineering to investigate the degraded VPN. The VPN services view overlays the service topology on a geographic map of the core network. Each node represents a provider or provider edge router participating in the VPN. We select the affected service VPN 201.55:58To open its detailed assurance view, provisioning succeeded, so the service was configured correctly through the controller and NSO.56:18However, the health state is degraded, indicating a runtime fault in the data or control plane. The service is monitored using an advanced gold tier assurance profile. This profile defines the key performance indicators and thresholds that determine service health.56:27Under the health tab, a single active symptom has been raised. An external BGP session to neighbor thirty.1.three.two is down on node three. This is the customer edge to provider edge peering that carries the VPN prefixes for the.56:48Because that peering is down routes are withdrawn and the service health degrades. The probe sessions tab reveals how the service is actively measured. Six probe sessions run continuously between node three, node four and node five.57:08These are Axian active probes, injecting synthetic traffic to measure the real user experience. Each session reports delay, jitter and packet loss between the endpoints. This is similar to standard active probing, but fully orchestrated by crosswalk. Hovering the monitoring settings reveals the thresholds behind.57:27Find the gold profile.57:47Interface packet error and discard thresholds are each set to ten. Processor and free memory thresholds guard the health of the underlying routers. Critically, the probe delay and jitter thresholds are each set to ten milliseconds.57:50If the axedient probes exceed these bounds, the service is flagged before customers are impacted. Next, we open the assurance graph to trace the degradation to its root cause. The assurance graph is a model that decomposes the service into measurable subservices.58:07The symptom details confirm the exact metric that failed on the device. The expression BGP session state equals up, evaluates to false. The observed value is BGP state idle, meaning the session never established.58:27This metric is collected per VRF for node three and its external neighbor. Model driven telemetry streams this state continuously rather than relying on slow polling. The full assurance graph now renders the service dependency tree.58:48At the top is the VPN 201 service decomposed into its contributing subservices. Transport summary validates the underlying segment routing paths between the endpoints. In a segment routing core, each router has a prefix SID.59:12And paths are steered with segment lists. The path SLA branch is measured directly by the Axian dynamic probe sessions on each node. These probe subsurfaces feed delay, jitter and loss indicators into the health computation.59:32The VRF reachability and BGP neighbor subservices validate the routing control plane. Following the degraded branch, the customer edge route health for node three is flagged. Its child, the external BGP neighbor health on node three.59:52Is the true root cause?01:00:12The graph cleanly separates the transport, routing and service layers for rapid fault isolation. This is the value of model based assurance, one symptom mapped to one actionable root cause. An operator can now remediate the BGP peering and watch the.01:00:16Observis returned to healthy.01:00:36Together, segment routing transport and active axedian probes deliver end to end SLA visibility. Finally, we return to the VPN services inventory. In summary, crosswalk unifies provisioning, segment routing transport and active service assurance.01:00:39This enables proactive closed loop operations for layer three VPN services. Thank you for watching.01:00:58Okay thank you yeah we are ending the demo session and thanks everyone for waiting for extended one or 2 min. Thank you. You said over to you. Sure.01:01:07Thank you everyone and thanks and as we open up the floor for the questions, please take note for additional resources. I'll just share the screen.01:01:22Okay, for additional resources maybe you can access our solution overview at the 1st link and you would highly encourage you to join Cisco automation.01:01:43Which is an active space where you can ask your technical questions to Cisco fellows and all other engineers and they share the best practices and kind of be collaborate each other with the all of our engineers. So we'll now address the questions if you have some.01:02:03You can post it into our Q and A panel and send your feedback as well as what you would like to see in the upcoming yeah we would be happy to include in that our upcoming sessions.01:02:23Yeah, sure. So it's been over 2 min actually our session. So as we don't have any questions as of now, we can close today's sessions. So you would receive a brief survey about this sessions and can just give your feedback.01:02:42Yeah, it's a short survey after 2 min and it would be really valuable for us to learn future sessions in the sease.01:03:02Thank you for your time and we'll see you forward to see you in the next sessions. Have a great day and bye bye everyone. Thank you.01:03:10