Nexus Dashboard Telemetry with advanced use cases

Oh, let me tell you a little bit about today's session, right? Today is about next dashboard in action. So we will walk you through real operation.00:00Scenarios, right? And we will try to connect it with each one of the features, right? And the goal is simple, right? The goal is to show you where in the.00:20Saves you time right where you can reduce noise, right? And how it can help you cut MTTR of your troubleshooting right and and and your production environment. So let me show you how the session is structured, right? So we have this flow of the of the session.00:29So we will go through three different phases. 1st one, show me how, right? This is a quick platform overview and the structural architectural shift, right? That happened. Then you have explore with us where we're gonna, this is where we go through the practical.00:50Scenarios right and map them with the features and workflows, right? And then help me succeed, which is a quick summary right quick Q and A and and one to one follow UPS for later on, right? So will be live during the session, right? So here's where, that's where.01:10You can ask questions or we're gonna ask.01:30Also questions, right? So, please, if you can join us here, I will leave it here a little bit just so that you can scan it and join us with Slido. Patrick, is there a way to see if.01:35Have people. Yeah. Yeah, alright. So I will probably leave it a little bit there so that we can join.01:54Let me see if your 1st question appears. So 1st question who can leave the audience now looking at it already? All right. So yeah, the 1st question, right? What is the biggest challenge on your fabric today? Let me.02:12Okay, with that out of the way, I will continue please join the the Slido, ask questions here for for us and we will try to respond, right? So with that, we go into.02:31The, the 1st slides, right? So.02:51This slide is all about four different pillars, right? I'm trying to unify our infrastructure, our network, and how it could look, right? So provision, right? Define ones and then you can deploy it anywhere. This includes things like automation, right? Infrastructure as a code.02:56Workflows and things like this. Then you have secure, right? So.03:16We need to have one source one of truth right for segmentation and, and zero trust, right, for our networks. And then you have managed, right? Which help us having this visibility instead of having isolated domains, right, different fabrics that I need to go and take a look at the.03:20Of them, right? We have everything on on one single pane of glass, right? So that you can work with that a lot better.03:40And you have analyze, right, which helps us identify issues, right? Or shoot faster and fix problems with more confidence with this, right? And the key message of of the platform right is is that ND is not just a reactive tool, right? It's something that you can use and you it can become proactive, right? It can be.03:50Some predictive as well depending on how we're using it and and how we're applying the.04:10The the different features that it has, right? So now let's, let me zoom in on on and show you a little bit about the broader strategy of this unified data center, right?04:17So you probably have been seeing this unified thing, right? So there are three columns here on the left unified architecture with AI ML fabrics, right? All these new stuff. Then you have in the center unified operations with MD.04:33And then, sorry, ok. And then you have unified intelligence, right? Where you have next dashboard but you have also integration with data broker, Splunk, thousandeyes, which is something that I was talking about a little bit in the, in the beginning, right?04:51So the the key point is that operations and intelligence sit on the on the same platform, right? Telemetry will come through the infrastructure, right? And then feed into our platform which becomes a usable operational insights, right? And we are able to use this.05:10This this data for our benefits.05:30So with that, let me now show you a little bit of the architectural change right that that it was very needed, right? So we have on the left the older versions, right? So three.13.two. And if you work with it, you know that we have a platform, right? So.05:33And you have the fabric controller, you have the NDO, you have the insights, right? And each of them have different cycles, right? So it was a bit of a headache to do an upgrade on having to go and take a look at the matrix, which one I do 1st, right? ND or do I do when NDO, you know? All these all these things.05:53That.06:13Became that gives a lot of friction rate on operational, so each of them had one single life cycle and and it was too hard, right? So now today you have next dashboard four.one and four.two and and on, right? It's one single unified.06:14Image, right? So automation, NDFC orchestration, everything comes into insights, everything comes into the same platform, right? This means that it's one single upgrade, one single API surface, right? One single dashboard, and.06:35Apic remain the same way, right? So it's still outside the two way integration that nothing changes at all in that, right? In practice, the unified version rate or or experience means that we will have.06:54Less windows, right? Less, less operational friction, it's a lot easier to maintain, a lot easier to, to do the upgrade, right?07:14So this is the architecture. Now let's take a look at a little bit about the fabrics, right, that ND can manage, right? And here, covers and Andy covers three different main fabric categories, right? So we have the land, we have ACI and and SAN, right?07:21In land you have things like Nexus Nexus catalysts, iOS xrs, 3rd party environments as well, right? Then you have, that's for NDFC. Then you have ACI with Nexus nine case policy based fabrics, right? So with Apex as well external.07:41And then you have SAN with MDS nine case, right? So everything running nine case over here on the sand with smart smart sounding as well. So across the three different, different fabrics, right? The value is consistency, it's.08:01The same interface, same APIs, right? Same operational model, licensing tier, right? It's selected per fabric, so not per per cluster of ND, right? So more per fabric for for.08:20ACI or or or NDFC, right? So now let's take a look at day two operational, right? And how ND comes here, right? And this easiest to read, right? That's four different questions that ND will respond, right? So for assurance, e.g., are we following the best practices and the.08:37There are a few more that you can do, right?08:57For troubleshooting, where is the issue and what is the blast radius, right? How is this affecting us and proactive advisories, right? So could we have prevented these, can we patch things vulnerabilities and such things, right? And you have single pane of glass.09:00You can, you can watch everything from single ND, right? So you can bring up all your, all your fabrics in inside one single cluster and that way seeing everything on on the same, on the same cluster, right? So ND addresses all these four, right? It it will mostly change on.09:18The.09:38Tier that you have for for licensing and then with that with the depth of of the features that that you get, right? So talking about the features, so over here you have a slide with all mostly all the features that, that we have, so you can see these have different three different buckets, right? So you have visibility amo.09:39Monitoring analytics and correlation and advisory and tools, right? And you can see the kind of the the right visibility topology capacity planning, connectivity analysis microberth, things like this. Then you have analytics will which will bring telemetry, assurance and point analytics, right? And I'm.09:59Anomalies and how these things.10:19Go together right for for the example. And then you have advisories which are more like reports, things, right? Conformance and life cycle pcrts, sustainability, you name it, right? Kind of more reports. So.10:25We're gonna touch most of these, not each of them, right? But, but they will come into different workflows or or things like that, right? I guess the point is, you have all these different features and not one single feature does everything, right? You will use different tools.10:42For different things, right? E.g., if you wanna.11:02You're working with your contracts, right? Maybe you're taking a look at the contracts, maybe you're gonna go for capacity planning, right? Maybe you're gonna go for the TCAM utilization, you're gonna take a look at that. Maybe you will do a connectivity analysis so that you can see if you're hitting or not the contracts, and, and, and then you get the point, right?11:05So you can use different tools to tackle different or troubleshoot different things, right? So over here we have the three use cases, right? So, 1st use case I will try to go short on that just visibility and and custom dashboards, right? This is the point of kind of.11:25But I would look at when I bring up or when I start my day, right? Then we're gonna go for troubleshooting connectivity issues, right? And then.11:45Design and best and planning with pre change analysis and delta, right? So I will be presenting the 1st the the 1st use case and then my colleague Vincentur will help me with the other two, the other two.11:56Use cases.12:17So, the way that we will do it is we're gonna introduce 1st feature then we're gonna go for the, for the use case, right? So on this slide, let me try to do a little zoom in there. On this slide you're looking to a multiple.12:18Of ACI topology over here, right? So you have.12:38Leave one of four selected and one single important thing right is that you can see most things, right? You can see the health of this, you can see the remote leaf status, you can see the IPN status connections, BP neighbors, uplinks, just one single link, right? And taking a look at that.12:42Replaces a lot of the manual checks and and CLI commands right having to go one one to one just so that we can have a quick, a quick.13:02Check or also will include things like health, right? So that we can take a look at the, at the color and just with the color we can, we can tell if something is broken or something is not broken, right? The topology updates, of course, automatically we don't have to be clicking things and this can also help for things or tasks like VPC perioding or.13:11Will be really right maybe checks for NDFC, e.g..13:31And one of the most useful parts, right, of course in in in MDFC is the historical comparison that you can go through and you can go back and you can compare different snapshots, right, and quickly answer questions like what changed overnight if did anything changed, right? And who get that changed so.13:38Helps us a lot on that part, so once that looks good, right? The next point would be would be go to our next dashboard, right? I mean to the custom dashboard, which.13:57We can build, right? So here you have the the next to-the dashboard creation wizard, right? And as you can see, you can start with different templates, you can build your own. The important thing on this is.14:13Making the dashboard for the right audience, right? You can create different dashboards and assign to different use users, right? So that they can have these different views, right? E.g., knock NOC will need more real time anomalies, right? They will probably.14:29Really want to see the health of the fabric, the anomalies of of the of the switches, right, the interfaces et cetera, and points you name it. Architects may need more for trends, right? Trends and capacity so they can plan better expansion or scalability of the, of the network, right? And you.14:49Have things for management, right? Maybe you wanna show them or we wanna have a clean report or a clean summary for, for the, for, for this in one single glance, right? And you can precept global home pages to have or.15:09Have something like this like ready for exec executive consumption, e.g., right? So with that, let me, we jump into the 1st use case, right? Which is custom visibility, right? And we're gonna go through those.15:27Yep, so the 1st thing and and I guess the point right that I want to show is is that NB.15:47It Helps you take a look at the bigger picture, right? So the 1st thing that we're gonna see right when we log into it is gonna be this welcome or this dashboard, right? This dashboard is at.15:54My cluster level, right? So everything that is in my cluster, if you can see here, we have two fabrics, right? So it's my both fabrics in one single glance, right? The important things to see here is the, e.g., we have the anomaly level, we have advisory level and very fast we can see that zero has occurred in the.16:10Last week, right? So even for for a lab, this is quite good, right? We have two criticals, but it's not so bad because this probably had been dragging for a while, so it's might it might be just maybe a port down or something like that, right? That that is getting there. But nothing changed last last night or.16:29Indeed that this week, right? Same with advisories, right? And then you have recently activity so we can take a look at who did this change. On all of this, you can click on them and it will.16:49Drilling so that we can take a look and see if something changed, right? So we can see that major did change on on on yesterday, right? And then you can start drilling. Advisories, right? Same thing. You can click in and and take a look at that. Then you have the infrastructure.17:05Sure you have the the right?17:25Then, I wanna go into my topology just to show you, right? So we have, we have here my topology of my two fabrics and we can start drilling in into one single fabric, right? So right now I jumped into my ACI fabric and we can take a look at one single lens that, ok, my spines are healthy, I can see.17:28The connections, I can see the connections on my leaf but this one does have a warning.17:47And this one critical, right? This one will have major. So we have the three flavors here or the four flavors here just to show you right and and show you how we could look. From here you can jump into a specific device, right? And zoom in into that and go out, right?17:53For now I'm gonna keep it like that. I'm gonna go into my dashboard just so that I can show you this template or or this show of this dashboard right where I can see my top.18:13Fabrics right so I will I I can have all my fabrics, I can have my devices. One thing that I mentioned right about the time selection and things like this, which come very, very useful in, in NDI, right? Is that you can go back or you can take a look or grab a good chunk of of time, right? Maybe last week.18:29Even or you can grab a time frame, right?18:49And then you can apply. So e.g., I don't know, let's do that. And you will see how it it will increase, right? So, from here you have the fabrics, you have the devices, you have everything that we wanted to see. We can also add more.18:55Or change the things that we wanna look into each of the of of of our of of our add ons in, in the dashboard, right? Okay.19:11So, now I wanna maybe take a look at my specific fabric, right? This is the same as me as me going into my managed fabrics and then going into my specific fabric, right? This will take a look at this whole data center or this ACI fabric, right? So now we're seeing.19:28The whole thing at a specific location, right? We can see kind of the same vibe, these all things will have kind of the same vibe, right? So you can see fast anomalies zero change advisory zero change. I can see my general information of this data center, right? So I can see the fabric name, I can see the.19:48ASN, I can see the license tier that this has and my version, right? One thing that I want to show you is the telemetry status, e.g., I can click on that and it will take me to the telemetry status. I can see that everything is running good.20:08I can then jump into my switches so that I can check that my assurance, my hardware search, everything is running well, right? So in assurance I'm having some issue right there but in terms of telemetry I am collecting telemetry on all my all my switches, right? So this is what I wanted to take a look.20:22We can see the recently activity. One thing that I like is here on the resources, hardware sources, e.g., we can click on that and you can see for the whole fabric.20:42By CPU, by memory, by the power supply, right? You can change it, you can change again the timing, right? So that you can see maybe more time in here, right? So you can see from 7th of April to 14th of April and you can then change either if you want to check maybe the temporarily.20:53Sure everything is running well, my find utilization, right? Everything is ok.21:13But you can see spikes on on the time, right? Where, where it comes from. Another thing, the capacity, right? It's very similar to things that you have in in ACI, but the cool thing over here, right, is also you have filters, right? So you can filter always.21:19In this kind of of views and of course you can you can change the the top switches, you can check check by which domain or different stuff here, right? Configuration resources, interface resources, and name it over here.21:39Let me go back over here and another cool thing that you can ask, right, e.g., from, from here you can check the conformance, the compliance traffic analytics, which is something that Vincent is gonna be showing you. So I'm gonna skip that one. And then you can see also the energy management.21:59Right you can enter here the the amount of the dollars per whatation then it will calculate it for you. So you can take a look at that.22:19Inventory so you can check all the switches, controllers in this fabric VPC pairs, right? Connectivity. This one is good, right? This one will let you see all the all the all the interfaces right in in your fabric, and talking about the the filters, right? This is where we can maybe take a look at or or use it, right? E.g., if I.22:27Wanted to use all or see all my 25 gigs that I have in my environment, I can do that and we will create a filter really quick for 25 gigs, right? Or if I wanted to look at one specific switch, maybe.22:47Like that, right? Then I can take a look at the management. So you can also create your or set up your own filters here, right? So if you just like it will give you options and it will help you create the different things. So if something contains something, you don't, you're not sure.23:01About the naming, right? That is not the exact name, always use contain e.g., right? Another thing that you can take a look at this connectivity thing, so you can take a look at L three neighbors, endpoints.23:21Right, maybe I wanna take a look at some specific thing, right? So we are drilling in into all the way to a interface, right? We can, we can do that. So right now we're on the on the leaf.23:33From the leaf we can take a look. Same field and look, right? Anomalies, advisories that you can take a look who was the last one logging into this switch. I can see this switch view, right? You can change it as well, but over here we will let you see quickly.23:49If ok so if it has a name, it's my apex, right? So you can see how it will categorize and show you really quick some if something is wrong, right? And then you can drill in into.24:06These issues and take a look at different things, right? E.g., now we're on on this interface and then I can take a look, e.g., at my, my utilization rate on bandwidth right? Anyways, so you can see how the.24:22The NDI can come, become right a nice tool to.24:42Going, take a look, make sure that nothing's is happening, right? The morning routine can be very quick, right? We went through a rabbit hole and start investigating things just to show you how you can go in and out and see the big picture, right? But the biggest morning routine would be just coming.24:49Been seeing that everything is well is well same as yesterday, right? In my topology, everything is good my dashboard.25:08My reports and we're good to go, right? If something changed, then we can drill in and start working with that. So with this the morning routine is done and let me continue a little bit about with the different features over here. So let me.25:15Zoom in a little bit. Alright, alright so ok so here's where ND becomes also practical, right? We were talking about the anomalies and all thing in this example.25:35The anomalies interface down, right? An event. An anomalies is kind of.25:55Mixture of faults with, with, with events right that that become correlative and then becomes an anomaly, right? With that NDI will show you, right? And in this example, this anomaly is a event down of an interface of Spain.26:01Of a spine rate.26:21In the answers two immediate questions, right? What triggered this? So this spine is what triggered the the interface 01:30 three.two, right? And then what is the impact, right? So you can see that the impact has been impacted one in point be lands, right? And six.26:24Additional correlative and anomalies. These correlated anomalies are the.26:44The real value here, right? The correction tree will show you these things that that is not just a flat list of alarms, right? That everything's red and how do I tackle it? At least it gives you a little root, right? So that we can.26:49Tackle that one and if we tackle that one, most of them are gonna fix downstream, right? The downstream effect, you will see it and then we can start tackling these others if there are others, right? So that means that there's less noise, right? You can take a faster action into into the.27:07And fix it and and fix the child alarms we did, right?27:26In one in one go. And then right ok then we go into the connectivity analysis. This one is kind of like a trace, right? These traces that are the traffic between two endpoints that you can set up.27:32You can sell, different things like ports. You can use this, e.g., to validate contracts for.27:51To validate that the path end to end is is going right on this slide routing is ok. Traffic is ok, security is ok, right? And the path is shown across two different data centers, right? So you can see even the path through ISM, which that's something some.27:59Thing new as well, right? And behind the scenes and the combines is combining multiple checks, right? So we do use Elam one right for consistently checks and and checking the traffic and the routing, and then it lets you visualize the path, right?28:19So with that connectivity analysis confirms the path, right? So now let's look what happens if we do have like an intermittent issue, right? Which then will take us to our second use case, right? So in this, in this slide, it's so.28:41All about turning raw data into usable insights, right? So this example, the overall the score critic the score is critical, right? You have the congestion with warning, you have the latency, so but.29:01You have drops which are healthy, right? So there's NO drops in in our in our in this specific traffic, we are seeing something is dragging the traffic or we're having lack, right? There's kind of a congestion.29:18But there's NO drop, so at least we're not losing any traffic on that on that matter, but we can take a look at what happens, right? So Andy also categorizes the traffic automatically with the traffic analytics with using the layer four information, right? And, and then this helps.29:33Just see which services are the most involved or most affected rate and then we can take a look at our click and this is very powerful when an app says or when when the team or they just give you a call right and tells you that something feels weird or something feels slow when they don't really know what and it's always.29:53The network fault until we're able to demonstrate right and show them that it's not our our end, right? Or if it is, we can fix it. So, that's kind of, what what I have here.30:13And now can can you send 2nd question, please, Patrick? Yep. I'm starting it now, yeah. Perfect. Thank you very much. All right.30:31And then we can go for the next use case. So I'm gonna pass the ball to Vincent here, let me Thank you Artro. So it looks like the problem is on me. I have to troubleshoot something. I'm NO more intact, but I will help you on.30:47That. So indeed we have a scenario here I think that we need to cover, which is basically a day in the life of a operational engineer, so.31:06Let me take back to the scenario here. So let's say we have a ticket received application connectivity issue. Someone is complaining that there's a problem with intermittent connectivity for a data analysis system and well we are lucky, we were given the IP address of that service. So.31:17It looks like this is something we need to look for timestamps. So since yesterday morning around 11:00 a.m.. The issue was observed and that was ok the worst case is not consistent problem is something is happening, these are bringing coming back, which is not something engineer liked to have so needing to figure out.31:36Basically what's happening because the connection is going and is is dropping. So the information I have is well basically usual questionnaire is what did happen? When did it happen? Where? Where I know, ok, 1702 1607. So let's start looking at.31:57That service, that specific maybe application or web server and I will skip everything before yesterday morning and the issue seems to be continuing. So, ok, let's let's have a look and let's see how nexus dashboard can help me on that.32:16So now I will jump to my lab, then let's let's try to, to get Andy help me on that. So now I'm on the main panel that's how to give you for information. We have two fabrics in this environment, so let's go into manage fabrics and people say that this is for this.32:32A specific endpoint that I know yeah that's ACI so that's.32:52Skip the next OS fabric and let's go into the ACI fabric. So ok this is my environment, the main dashboard, my inventory, and well, 1st thing I want to look because the tool is helping me is about anomalies. So it looks like I have some problem in this environment. Let's jump to anomalies, so.32:57This is what situation is as of now right now. You see, information is active now, which means this is lightest snapshot. We take every 2 h a snapshot of the fabric and then every few minutes again subsequent we receive telemetry from the fabric. I can jump back in time, so I was say that this.33:16Started.33:37Over last day at let's say eleven, so this looks since last day and now the filtering would happen that I will just see issues for that point in time till now or from yesterday. I can group the anomalies or un group them. So in this case, let's have a wide view and let's see about something ungrouped. I can acknowled.33:38Each problem I can assign them to someone, I can select specific anomalies, so and he's capable to do root cause analysis if you have correlation of problems, again, that helped me, for instance, word scenario, I have a routing problem, I have peering issues, but maybe this is not caused by the next router and I'm maybe start focusing on this.33:57Troubleshooting plan from my head and I will lose a lot of time. So root cause analysis maybe will tell me this is not the next hop, this is maybe local switch and maybe that's the sport and maybe that part has an SFP problem, so you see this kind of root cause analysis.34:17Can be done by the product you can tell you, well, don't jump to conclusion about routing protocol. I already found out that this port has some connectivity problem. So that's helpful. So in this case, let's keep all anomalies. Why they're interested is what started and what is ongoing. So the 1st two ones, I see it's detection time December 25, ok, that's.34:31So long time ago it's still an ongoing problem. Someone may have started to look at it or I can maybe assign it to someone. What I'm interested on is some connectivity problem and yeah, this started ok.34:511011 today it's ongoing. So it looks like I have things to look here. I have problems that are similar here so I can select one of these anomalies as my starting plan for my troubleshooting.35:02So, ok, I have issues here about egress utilization for that port 01:13. Bandwidth utilization is well high and 98 %. Flow drops from this source to that destination.35:17Well, the tool is helping you because you see 1702 1607 that's yeah that's curious and accidentally helpful because this is exactly what the people started to tell me. This destination, this service running on this IP address. Well, it looks like I have something here.35:33So let's jump to that one. So this is an anomaly, what this anomaly has to answer a few questions. That's mandatory to avoid false positive. So what is wrong? Well, we had a detection because we know we have telemetry ongoing, we have flow flow ongoing, so in this case, traffic analytic is using NetFlow V nine enriched. So the nexus fabric.35:49Is sending or the ACI fabric are sending this telemetry at subsequent, so you use PTP for that and that information is aggregated in ND and ND can correlate this information from.36:09The other statistics, the port information, this NMP traps, and then this this information from from NetFlow. Netflow is oriented on destination, so that's why my focus will be anyway to troubleshoot the providers, so 1702 1607.36:21But if I want to know more about the source, the th the let's say, the server or the host that wants to connect to that provider. So that consumer, let's say, that consumer zero three, if I want to get more information, then I can use this capabilities on Nexus switches, which is called.36:38Called flow telemetry and flow telemetry events. But 1st, let's look at destination. So for packet drops detected due to buffer drop, before max drop.37:00When did it happen? Well, it happened yesterday eleven. Again, I'm super lucky. This is the time when people started to complain. So this seems to be a match with my problem. I I started with the goods ends of the of the of the the problem.37:09When did it happen? Okay, so it happened at that moment where on leaf two, ok, I will have a look at leaf two later on, but what triggered this anomaly? So we had a flow from this source and indeed any random port to destination 8080 packet drop. What is the impact? At least one flow is impacted.37:26By this anomaly.37:47And I have as well two virtual machines affected by this anomaly. So it looks like 1702 1603 is one, but they are more people affected. So it looks like again you see anomaly is cleared? That's a flapping problem. So it's ongoing, it's going away, it's coming back, Seems to be that this this description of my problem.37:48So 1st thing 1st, ok, let's see which flow are impacted, which device ok so.38:08Which device are affected? I have two Vmiss in this case, source receiver, ok, affected entities. So I can look at that and maybe confirm with my, with the team that opened that ticket if these were the ones. And then let's look at this flow.38:13So, now I will go into my toolbox, so analyze. So we had anomalies advisories and analysis app is some kind of of a toolbox. So you see here connectivity are to explained to you, but what I want to do is traffic analytics. So let's jump and see fabric wise what is happening. So fabric is.38:29ACI two, ok? I can change scope to another fabric if required. Scope of time, current, ok? I can see what's happening now and what happened before. And again, we see traffic analytics reach.38:49Major problem over the last 2 h. I have congestion, I have drops. If I drill down, I see NO service categories and I can see all of the endpoints. What I'm interested on is basically more about my endpoints because I already know so service endpoint.39:01And then I see in this graph when I hit something above healthy, this is a deviation. I have major problem and then it falls away and then again and again. And then again you see the top one with this problem is 70.seven. So then I see some endpoints.39:17Here and I see again the 1st one is 72 1607 port 8080 again that match my description, and then I have some where is it connected? So it's in the demo tenant on VRF TG connected to leaf two, ok? Again, that's my next step, looking at what happens on leaf two, it's a major problem.39:35Category web indeed port 8080 is web server and then I see this statistics of transmit and receive of packets.39:55So let's see this one more in details. So I can filter again something I can see is endpoint details. So where is this server connected? Again, we know this is TG receiver, that's a virtual machine connected on this EASE.40:03Site two. How do I know that? Because I enabled the connectivity with the VMware vCenter, so nexus dashboard receive information from the vCenter about VM specific information. So again, added value, I can do integration of 3rd party products like VMware.40:23And other visualization platforms for instance. So I see this one was yeah that's the Mac address of this point. So if I need to take some note I don't need to drill down on another tool to sort out information about endpoint.40:45Last updated, so last statistic from a few minutes ago it's active learned. So that means this endpoint is in the fabric still. Again, information about my logic, so it's that VRF we know and have a bit more of details, so that's this bridge domain connected on this EPG and so if I need to troubleshoot config I can do.40:59From here. And then again, it's connected on on at least two leafs, so it seems what's leaf.41:19Two and leaf one. Now I have information about leaf one course one/one/thirty. Endpoint history, did it flap? Did it do V motion? What happened about this state of this endpoint, so you see, yeah, it was.41:26Moving from leaf two to leaf one, so maybe have some V motion, maybe I have some problem that VM is, is misbehaving, maybe control trigger. So that's on my list of to do of to do things possibly. I can add this information anomaly change indeed because we have drops. So anomalies related.41:41Nothing specific what I'm interested is my traffics. So traffic analytics, I can have more details about this another one, so I see not the the client connected to the four clients, and then trends and statistics.42:01You see these are the, the, the metrics about this endpoint. So I see the amount of connection, the amount of conversation from the clients to that server, the latency, and as you see, I have some very weird graphs here. It's high ticks rate, nothing, high ticks rate burst. So there is something bad again, I know.42:18What I want to look now is flow. So all I can do is trigger a flow collection. So I want to know more about the the Vmiss or the server that are.42:38Reaching this specific service. So now I have my information from netflow. I know that this destination is, is badly affected. Maybe I want to know more about that one. So what I can do is create some traffic analytics information and then create some flow telemetry information out of it. So this is something I can do for a certain.42:54Some point of time, so 3 min 5 min for instance I can capture more details from the flow telemetry capability, the hardware telemetry from the switch. So this is something I already have because it takes some time to run.43:14So traffic analytics, flow collections, and you see here view records. So when I do flow view records, I see, you know, the perspective from the initiator. So let's take one for instance.43:47So this is zero.three trying to reach zero.seven, ok, that's one flow of interest. Let's look at that record. So this is latency information packet drops, so we had a lot of drops. When did it start to capture information? So that ran for 7 min and we have this information about source and the.44:03So ingress, this is my client and this is my target. So my target I already had a lot of info. This is something new for me, ok, this is one of the client affected and again where is it connected? Leaf one port 01:11, but let's jump to this Visualization is helpful.44:24So now I see source zero.three connected to leaf one goes to spine down to leaf two and leaf two destination is connected there. What I see here is point of interest color code. So if I click on the switch, then I see information about that switch so.44:39Ingress leaf switch one, so it start when the source is connected to port 01:11. Then we know next top is the spine one. Local port 14053 spine port one one ok we don't have anything there specific healthy traffic again on the spine so the spine is not affecting us, but leaf two is my destination.44:55So if again like click on the, on the ISO link, then I see.45:14The latency there, each port on each end of the of the ISL and then I see leaf two is of interest because leaf two is in red. So here what happens? Why is it in red? Because I have buffer drops. Okay, I have buffer drops on egress, and again this is not traffic from spine.45:20To local port because that one is green, but it's egress from local port to destination. So on leaf two port one/thirteen, I have buffer drop. That was already on top of my head, but nexus dashboard confirmed that to me based on the information I gathered from the graph, from the counters, but this is.45:38Visualization. So again endpoint information ahead. So my next step is to look at this port one/thirteen on leaf two. So let's have a look there. Again anomalies that is wrong, so flow packet drop is sending me more information. So let's go back to the anomaly and then leaf two.45:55I can open front panel port vision of leaf two. So I can go again in many ways into the interface. I can go in inventory and then open information about that one, but this is leaf two, and I know I have to check at something.46:14Anomaly level again it's high because it's derivated from that port problems and I see you switch anomalies only, ok? That was flapping, but I know I have to check at port number 13. Or current time or last.46:30Day information so again I can check there. So port is up. I can see the endpoint connected there, trend and statistics. Again in real time, NO, maybe I don't have drops anymore, but you see I have this utilization that's.46:46Yeah, at least transmit rate almost one gig and it's, you see it's capped at one gig here, which is not something I would expect and see microburst around the same time as my peaks. So they are bursts, they are problems of queue so.47:00Well, maybe I need to do some QoS tuning here, but then it would happen all the time. So let's have a look at this one again and about that port, maybe has something to change either in the QoS or something else.47:17And something that brings Bell to me is this one, operational speed. That was kept at one gig. All the time I had problem I was using 100 % of the bandwidth utilization. So possibly I made a mistake here and one of my policy was to put that port at.47:34One gig while physically it can go ten gig. So multiple option I can have to fix that and again that's part of the anomaly remediation option that the nexus dashboard is proposing to me is enhance the amount of ports, so I have more distribution of the traffic to more ports or maybe I need to put that port.47:52About ten or 25 gig. So I have dropped due to congestion. So at some point I need to fix that congestion and this is proven by, by let's say this demonstration that I figured out what was the problem. I figured out where and when it happens.48:12And then I can possibly use an action plan to remove that issue by changing my configuration on that port. And in my opinion, the 1st and quick thing to do is look at my ACA configuration and see if I need to just change a policy to ten gig, and then maybe I've fixed my problem. But at least now I've documented. So.48:30Yes, this time the problem was the network. Oh, I think I'm done with that part so back to you Arturo if you want to demo. Oh, ok, let me bring out the presentation over here. Cool, thanks so much.48:49So yeah, I get I guess you know guys there there's NO straight up workflow, right? The thing is to look.49:10Add the big picture and then start figuring out how to attack it or where it's better for you, right? Maybe from anomalies, maybe, start drilling in into that, right? And there's many ways to, to, to go with with next dashboard, right? So.49:19Coming back into features, right? We have here software and hardware conformance, right? This is one of the part of highest value rate of operational checks, right? This answers, three risk questions, right? And are we running any hardware, any, any.49:37Software out of life or are we hitting any verified scalability, right? In our, in our network? The nice part is that indeed presents this as an exe executive friendly way, right? And we can really quickly take a look and get out.49:57Right if if there's nothing that we need to do, right? You can see current critical devices, feature risks, trends, right? Makes it easier just to, to.50:17Keep going with the remediation right plan. So one of the cool things that I like as the TCAM TCAM analyzer, right? So this you know Tcam is finite, right? We use it for everything, every contract, every policy object consumes.50:28Said, right? You know, any contract will consume at least two, right? Depending on how how you do things.50:48If you hit a ceiling, then behavior becomes unstable, right? For our switches. What happens is that it doesn't have space for it to to save the policies and then things become unpredictable, right? Because we don't know which one is is getting is getting there or not, right? Well, if we're at that point.50:55We we're already.51:15In in in trouble, right? But the policy camp helps you see utilization right for contracts, for security groups and per switch, e.g., right? And this gives you things like hit counts over time, right?51:17This means that you can separate active rules from stale rules, right? If something has zero hits in one month, then maybe it's better to make a note and then take a look if we really need it, right? And then with that we can use something like maybe the pre.51:35Change or the delta the pre change analysis, right? Where we can take a look at and change ok what happens if I, if I move this, right? But 1st, let's take a look at the delta, right? And Delta is very simple feature, it compares two different snapshots we can select.51:55And then, of course, it's something simple, right? It's just that a delta that will do between the two snapshots or something that also Apic has right kind of, but over here, the cool thing is that it will show it to you in this in in.52:12With this dashboard right where we can quickly see and quickly jump into the difference if we see that, ok, some critical appear then we can click onto into that critical one and it will show us, right? Without, and it will help us maybe.52:30See what changed right and compare things or resources such as endpoints, bridge domains, EPGs routes, right, policies. So it comes, becomes good, right? For comparisons, maybe when we do a change, maybe when we do upgrade, right? We do an upgrade, we do a delta 1st and then we do.52:47Later and we compare them, we'll check if something's wrong, right? Very quickly if there's a down on EPGs, then we need to do something right because we're having connectivity issues.53:08So, who changed it, what changes and when, right? This is something that also has, right? If you see over here, it has policy policy delta, right? This is the one that it will turn, it will turn things right and it will show you evidence that sometime some.53:20They did a change or what changed right and who did change when, right? So it helps us a little bit for that kind of things, right? And then we have the pre change, right? This is one of the strongest differentiators right in the highest tier, right, of, of, of, of licensing. The.53:40Pre change analysis let you simulate, right? So e.g., I was talking about the maybe.54:00Checking the contracts, right? So maybe you can take a look at those contracts that you made notes on and and maybe simulate what happens if I delete them, right? Will you show me it should show you, right? The simulation and if something is broken.54:07Or something comes up, right? And then we can really make sure that this config is not really doing nothing or or yes, right? So we can verify it that way without doing the actual change in production, right? It's not something that we wanna be testing on production, but we can do it.54:26With this imulation, you can provide the change with object inputs such as XML or JSON or you can also select it, right? So it's very simple. You can just select, ok, I wanna change the contract and then you can just type the name of the contract. It will show it to you.54:46So this becomes evidence based risks assessment for the deployment, e.g., right? And with that, we're gonna go and I'm gonna pass the ball again to my colleague Vincent. Yeah, Thanks so this time.55:04And.55:24We will again go into the same lab and I will try to avoid breaking things. So let's use this pre change analysis. So pre change analysis as told you, a simulation of a change. So before it happened, well, what I can do is prepare it the best way. It can be as well helpful for me to document that to my change manager. So if I am in IT environment.55:25The cab meeting then I can already prepare that for impact analysis of a change and just send that information as a report for the cab management that they decide if I'm allowed to do it in in production later on. So in this situation I have a scenario, so pre change analysis version one. That's the one I will do 1st and then we'll look at another one. So.55:45It's already running in the, in the background because it takes some time. So in the sake of we are almost at the end of the session, we ran that analysis in advance. So what I did basically, I created a simulation of.56:06An ACI change, based on the file. So I went onto my my apic. I prepared my configuration and before committing the configuration, I exported that into an XML file. So I did not commit my change. It's ready in the apic. And I came back into this one and I just did add analysis based on file.56:18I can also use.56:38A simulation for a very simple change over a wizard, so I can say a crud. So modify delete or add a configuration and I can say for instance, create a contract or remove a bridge domain, and then I can use this wizard to select which object in my existing ACI environment from the managed objectory will be affect.56:39By that and then I can simulate. In this case here, I have more than one object to do, so I decided to use an XML file that I just grabbed from the Apic before committing. So that XML file, I can also download it from here if I want to review it again and then I can open it if if I want to, to see the details. So.57:00Let's jump back to the simulation itself, so that ran already pre change analysis based on the epoch, so that was done this morning at 08:44 based on the snapshot of the fabric, so the configuration was like that and if I want to do this change, what will happen? So analysis was done.57:20So we look at delta and if I execute that change, this is what will happen. Before I had four anomalies, if I do my change, I have nine anomalies, but looks like I did not prepare well the impact, so five new problems. If I click on them, this is what will happen if I commit my change. I will have configuration compliance configuration.57:37Communication problem, I will have all up in VLANs and so on and so on. So better to prepare that in a different way.57:57If I look at this anomaly overlapping subnet, ok, double click on this one. This is something that may happen and well if I did my configuration, it looks like I'm leaking a/twenty eight, and I'm overlapping that with another one. So maybe I need to change my, my subnet scope. So recommended solution as in every.58:03Anomaly, this is what the Cisco best practices knowledge base and all the 25 plus years of tech information is into this product. Well, maybe I need to change.58:23This configuration the way it's recommended here. Okay so overlapping VLAN is something I need to fix from my preparation. Let's close this one, overlapping so look contract configuring correctly, so again I have a contract. If I do this.58:34Yeah, I may have contracts configuring correctly, NO value contract to prevent communication between EPG. Ah, ok, so I changed even worse, I changed my BD let's say communication type. Communication compliance. Ah, I had some compliance rules, so in advance I can decide what is.58:51Allowed and not allowed in my environment globally. So Nexus dashboard will not prevent me to do that, but in this case, as you see, it will warn me that I'm breaking a compliance. So communication compliance, if I do this.59:11This communication will happen. So my rule is segmentation PCI, ok, so these two EPG must never communicate in that tenant because I created the rule against that. If I do my change, I'm breaking my role.59:22Because I'm allowing this to HRMS FEPG to communicate to another one. So I'm breaking a compliance rule in both directions. So my contract that I created is also something against company rules. And then, yeah, configuration and compliance rule.59:40Maybe I made another mistake, yeah, I have some complaints rule like maybe naming convention. Yeah, that's the one here. Compliance rule naming convention. I created an object with the wrong name. I need to set a specific attribute or specific prefix. Okay I did a bad preparation, but at least I did not broke anything in real life, so I know that. Okay.59:57Yeah, but it's based on this.01:00:17Let's say conclusion and recommendation. Let's say I changed my plans, I went back into my big, I fixed my preparation and I did something better. So I go back into my pre change analysis. This time I upload again my XML file with my corrected configuration and let's see what happens if I do it again so.01:00:19I have now 2nd time pre change validation, so V two of my.01:00:39Change. And this time ok so same XML file with the new details. Delta analysis looks like better now. I don't have more anomalies. I'm not fixing existing one, but you see I don't have more anomalies, so maybe then if I check at my compliance analysis, I'm not breaking my compliance rule for any.01:00:44Of my tenant configuration best practices of naming convention.01:01:04So you see these are the one that are enforced. So this PCI segmentation, naming convention configuration. So in this case I'm not breaking compliance either. So I'm not breaking overlapping and that's the worst situation in, in this case and in terms of configuration I'm better, so I can.01:01:09Let's say with confidence, send that report to my cab manager and then I expect that I will have a green light to execute that change in my change window. So this is really helpful pre change analysis tool, only available with ACI, so not with any.01:01:28Us as of today.01:01:48Delta analysis can be done when I execute or I did my plan and I execute my change, then I can run a delta after as a post check validation. Delta is possible for NXOS. So next NXOS you need to be better prepared, but you still can assess the impact of your change.01:01:49All of that, you can have it in a CICD pipeline. So this is what I did. I did prepare my change. I validated, I fixed my problems, I run the change, I did delta analysis after and then I can access to the entire team that the network is.01:02:09Validate it, then they can just check now the applications and the services. I can put that again if I have new anomalies, I can put that in a plan to remediate maybe some, some issues with my network. And I can put that with API calls into a CICD pipeline. So I can automate my changes.01:02:27In let's say robots tools or Nexus as code, so you can use nexus dashboard in this pipeline for pre change validation, you send that information, is it green or red about the amount of anomalies you trigger? Then you receive that maybe in a Webex room or in another tool. Then you just click validate and it continues the pipe.01:02:45Another interesting feature in nexus dashboard is the explorer. So you can still ask question to the tool. So show can, so for instance, show endpoint and then you put the information about an endpoint, and that tool will let's say construct your question.01:03:07Based on the managed objectory information it discovered from the so show me endpoint and then it will jump you to the endpoint information or can EPG one talk to EPG two? That one will just check if there is a contract. So this is not as deep as connectivity analytics where we do Elam capture and we will the.01:03:23Real aspect of communication. In this case it's just checking. Is there a contract, so can EPG one talk to EPGB? Yes or NO? And then it can show you this is the contract and then you can drill down later on about details about that one. So Explorer is that way, it's not an AI yet. And that's scope at one fabric level only at the moment.01:03:43But that's sometime useful to jump directly to a panel because we have so many ways to do the same, let's say workflow. This case maybe you can avoidation and clicks in the interface. In terms of licensing, most of the thing we demonstrated today require Prometheus, so next is dashboard itself.01:04:02It's not the license tool. What is licensed is the fabric you're managing. So the Nexus or NXOS or ACI.01:04:22Fabric needs to be with all switches registered at the same level of or tiering of licensing, and the tool has this dashboard as well so you can see which feature, which use case, which capability requires which license tier. So if you have an ACI fabric running at advantage, for instance, you can run most of the use cases but not.01:04:29Telemetry oriented ones. So the connectivity analytics, the traffic analytics.01:04:49The pre change analysis that one requires but already if you have essentials, you can do a lot of things. So you see, you can already do inventory and provisioning information. If you have an XOS, you can already do software image management because it's unified, NO, so nexus dashboard we've shown only the scope of telemetry or insights today but that tool can help.01:04:53Also do previous provisioning of configuration on NXOS.01:05:13So if you have an ACI multi pod or multi site, your IPN devices or ISN Nexus devices can be onboarded as devices handled by Nexus dashboard at NXOS level. So you cannot do also do telemetry information but can you can also do provisioning and configuration, software image management so ISSU or districtive upgrading.01:05:17If you prefer. We also have the information about.01:05:37Compliance advisories, bug scan, basic one advanced use case. So the key message is every Nexus customer is basically entitled to use Nexus dashboard. And then you can use the feature that are in line with your licensing. So then that's nothing else than just deploying this tool aside your existing fabric.01:05:40Virtually or physically in terms of form factors. And again, you can change the telemetry to flow telemetry or traffic analytic depending on the operating system version of the switch.01:06:00But only the most recent ones support this netflow based telemetry which we call the traffic analytics. Earlier version supported only flow telemetry. But again, you can do that, you can evaluate that doesn't have let's say affect the performance of the suite because we gather that telemetry in band. Last thing is.01:06:11You can evaluate a feature so you can test in premiere. And if you feel like this use case is valuable for you, then you have a good reason to uplift your your licensing tier. You don't need to ask Cisco for an evaluation license of Nexus dashboard for Premiere because you you are expected to test and then align your licensing when you go.01:06:31Production grade with your formeric.01:06:51So as I said, you can change that at any time with your your existing environment. You can click from essential to advantage or premiere, and that will just enable the feature in the UI, otherwise it's just grayed out. And then if you need to say, I will stay with advantage for instance, you move back to advantage. There is NO specific reconfiguration required. You just tick a box and then you have to.01:06:57The capability.01:07:17So I think we are close to the end of time and yeah, last slide of question possibly can help us fill up this will be a survey and, and also if you need one to one consultations you can request them through there and then we will reach out to you guys.01:07:21Yeah, I've seen one question about SAN. That's a different beast because SAN is fiber channel and this session and most of the session are about Ethernet switches nexus. If you need something about MDS 9000 and the sun, again, reach out to us. I left my email because I handled the land at the and the sun lab, so you can.01:07:40And have a session specifically on SAN. We did last year a webinar on SAN, but if you need a one to one.01:08:00Feel free to reach out and then we can go in details about some analytics which is the equivalent of inter this insights on on Nexus and maybe look at the use cases and your specific environments if it's open system or we have, we have also a good scenario and demonstration environment for that.01:08:06Yeah, if you need more information, I will still have a community.cisco.com when you can ask extra questions after this session and yeah, we can have knowledge sharing across peers, but as well as Cisco employee usually look as well in this place to give you an answer if.01:08:26The community did not find it earlier. So as we advised earlier in the session, we also have other topics to cover in next couple of weeks, which will be integrating with so ABPF best practices on ACI and visibility and observability information.01:08:45With Splunk. So these two sessions are planned for next couple of weeks. And for that, thank you for your time and your attention today.01:09:06