Beyond Visibility: The Age of Intelligent Assurance with Cisco

Um, hello everyone. Good afternoon, good evening. It's probably towards the late evening and really appreciate you all being with us here.

Uh, my name is Nikita Shahida and I'm one of the leaders on the product management side on the assurance team here at Cisco. First of all, I am, and I know y'all have heard this from many people, but I'm very excited to be presenting in this room today, and I'm sure everyone probably starts off, but I'm really excited. The reason is, uh, also I think the last time I presented to all of y'all was at TFT, uh, which at, during pandemic, sorry, this was like in the virtual session, so we couldn't be in the same room, but just having to be in the same room, I think just like makes a lot of difference.

So really excited and hopefully I can continue the excitement with our topic today, which is beyond visibility, the age of intelligent assurance. Now, you all must be wondering why beyond visibility, Nikita, because I know there's a lot of, uh, you know, rumors out there that we just do visibility and not beyond that. But I do want to kind of like put that to rest and be like, we are baking in a lot of intelligence into the platform.

Uh, we are doing more than just visibility. We want to be able to do actionable insights, and we are bringing, like I said, intelligence into the platform as well. So hopefully I can, uh, you know, uh, convince you all today that we are not just doing visibility, but we're not, we are not just stopping there, we're doing visibility and more so with that, like, I wanna kind of like set the stage with the vision, right?

Like the vision for Cisco is to bring or unify experiences together. Now we have our catalyst, we have our Meraki, and we have thousand eyes, right? There are like all great assurance products that have like really great PO capabilities and feature sets.

Now, what we wanna do is we want to unify these experiences together, and we wanna make sure that we give, we build more smarter assurance capabilities that are end-to-end because each of them have like their own strengths and we wanna bring them together and make sure that like, there is one cohesive story for our Cisco customers. So no matter where you are connecting either at home branch remote, anywhere, you're, you are capability for the IT admin or the network administrator to troubleshoot a network should look and feel the same. So that's where we want to get to, no matter what kind of Cisco capabilities you're using.

So this is at a vision. There's some pretty vast differences in your platforms today, and especially around this, this sort of data, right? You said you're, you're trying to bring it all together and one it does, does one win out.

Does, does like assurance today in the Meraki dashboard look like what it does in Catalyst Center? Like which one wins? Because they're very, very different products right now.

Yeah, I wouldn't actually say like there is a clear winner, but what I can tell is it depends really upon the person or like the organization and what they're looking for. We have customers who really like going deep and like, like the complexity really like technical language on the dashboard. Whereas some customers like the simplicity of the dashboard, but at least hearing from Cisco Live and a lot of customers, they want it to be more simpler, right?

Like we are moving into the age of intelligence where it's gonna just simplify a lot of the complexity. So I think it really is, you know, depends on customer to customer, but I think what we wanna do as Cisco is like, we wanna make sure that simplicity is at the core of network assurance. So we wanna bring a lot of like, you know, capabilities in the most, or give our customers capabilities in the most simplest form.

So there is like a lot of like the Meraki flows that you're seeing is what you will probably see that kind of lead, that experience. But there's strengths within each portfolio, which is like the path analy, the path visualization that you see on thousand eyes is really popular as well. So we are picking segments that are really popular across the platform and making sure that our customers get the best of, you know, that's there within the Cisco portfolio.

And, and that's gonna be a unified, uh, assurance score regardless of the platform or like, Correct. So we are, we are not there yet, but we are going to get to a place where there is, you know, a a same score no matter which platform we're using. Okay, got it.

Yeah, so it has to be like when you're using hybrid, when you cross po like you, you cross go, go into like another interface. The score should remain the same, but if you're just using one platform, you should still be able to get the same score as well. So that's, that's where we are like heading towards.

Great, thank you. So, um, while that's the vision, we want to be able to make sure that there are like milestones that we set in order to get to the assurance story, right? This is not net new y I've heard this over and over again.

Now what we wanna do is like with like baseline and detecting, we have certain like ways that we have like understood what thresholds are across the network. We wanna be able to, like you, I I'll be talking more about it as well. So just quickly going over this, we wanna baseline and detect, we wanna localize and diagnose what the problem is and when we say localize and diagnose, we would, we are what we, what I mean is bringing contextual information to the particular client that you're trying to troubleshoot to the particular device that you're trying to troubleshoot, to the particular network you're trying to troubleshoot or the organization.

It really depends on the context that you're trying to like really look at and we wanna be able to localize and diagnose the problem then soon mitigate and remediate. What we've done here is we've built platforms and frameworks that no matter what kind of issues that you run into or issues that you run into, maybe one or many, this framework will be able to kind of like put these data together with intelligence and like tell you exactly what you need to do in order to remediate it. I'll again talk about this as well and obviously put it and optimize with more data with the intelligence, we can only get better predictions and that is where we wanna be at the end state.

So this is, uh, the next one is a quick glance into some of the capabilities and features that we have in each of these stages. Again, this is not an exhaustive list. There is a lot of like capabilities and probably I will need more time and probably two or three more slides talking about it as well.

But I have curated this because I wanna keep our conversation here more meaningful and more targeted. So I've curated this list, but what I really wanna show is that we have investments in each of these stages, right? I started off by saying it is not just visibility and this is where I wanna show that we are not just looking at the first two stages.

We have made a lot of improvements across these different stages and there's like, there's more to come, but again, it's a very short list for us to kind of have this conversation. Previously, your story included, um, hardware sensors Yes. Which are no longer on this slide.

Um, is there, is there something that we expect to be filling that gap at some point? Yes, we are definitely working towards it and probably, uh, you know, I cannot speak much to it. Uh, but yes, that is something that we're looking into and hopefully we'll be able to talk about it soon as well.

Thank you. Yeah, but in the terms of sensor, to answer your question, we have synthetic testing and you can see the asterisk there because obviously there's more to come. We have the thousand eyes capability, which is our sensor capability within, you know, um, the Cisco portfolio and that's there on our WAN devices or our secure WAN devices right now.

So, um, we wanna bring that more into our, uh, you know, switching and our wireless as well, uh, which we're working in. Yeah. Um, so yeah, some of the things that I've like highlighted in the orange boxes are the ones that I will be going deep into.

The reason why I chose them is because we announced this at Cisco Live, um, this year or like just today, this morning at the keynote. And, uh, this was a part of the keynote announcements as well. So I do wanna kind of like dive deep into it and like show you what are some of the capabilities and why we did what we did.

But hopefully this kind of like shows you that we are not stuck with just visibility and I want, I hopefully I've put those rumors to rest right now and like we've done a lot of things, uh, which k kind of gets into the predict and optimize state as well. So what I wanna be doing is a little dance between a slight show and a demo. So to kind of like give you, you know, a one-to-one comparison of what I'm talking versus like what the, you know, the feature set looks like on the dashboard.

So the first one is a org wide assurance visibility. Now why is this important? We have a lot of customers that have like large networks, right?

They do want, they do want a quick analysis of what the health of the network looks like. We, we have constantly had customers saying like, Hey, at least if you don't have this visibility on dashboard and if you cannot build it for us, give us the APIs, but now this is available for them. So no matter you have a hundred networks, a thousand networks, this will quickly be able to tell you exactly what is the criticality of the networks and what are the networks you need to look at it right now.

And all of this is based on the score formula, which I'll be talking about as well. So that's kind of like a way for us to determine which networks are poor, which networks are good, and how can we really get the user's perspective to, you know, troubleshoot the criticality of the network right away. Has this gone ga yet or is it still considered beta?

So this, this has gone ga like as of like right now. Right now. So you will, we are slowly rolling out, uh, so we probably won't have it like for all our customers, but it is ga people are getting it.

It is, people are getting it slowly, yes. So let me quickly show you what that looks like with the demo. So right now you can see that this is our corp network.

You can see that there is a WAN appliance that is having a problem, but it tells you like quickly the device status, right? And this brings into perspective the devices across all your networks. We have seven by 84 that are probably having a, uh, an issue right now out of which two of them are poor, five of them are fair, right?

So it kind of tells you like, okay, you know, I do have some, uh, networks that are having a problem. And then if you scroll down even further, it tells you like what are the health scores, right? I'll talk about the scores as well.

But the health scores kind of tell you there's 46 points and the reason being this network device has 10 points. Now, network device, kind of like, you know, one thing I'll be going deeper too is like we've divided the score or divided the performance across the network by client's, network devices, infrastructure applications. These are the core components of your assurance end to end.

So now that you can see that there's network devices having a problem, this is a network that probably has the WAN appliance that is done. So really contextualizing and bringing these information together where you don't have to like look and fee, look at different pages is really helpful to understand that you know, what is the health of your network as soon as you land on this page. So What kind of application data are you gonna be tempering this score with?

Is that is like Salesforce application or internal applications? Can I write my own custom defined or Yeah, good question. So these applications scores are coming from thousand IG Okay, From thousand eyes.

Yes. So these are all, So you're hooking into the, the the thousand eyes dashboard or there's you're using that? Yes.

Yes. And I will show that as well, uh, during our demo. So we are hooking that up.

So the thousand eyes, obviously the reason being like we have the synthetic capability on our WAN appliances, but we, when we bring this into our, uh, you know, the access layer, there'll be the scores will get even better. So that is what we aim to do. So this is like a quick understanding of like how, you know, your, you know, just this is a NOC view so customers can put this up, have a NOC view and kind of like, you know, get an understanding quickly across like different the health of different networks.

Um, so the other thing obviously like, you know, going back to like say if I wanna look at like San Francisco, right? So you can really, really look at the San Francisco network over here. So once you expand this, it tells you like quickly, like what is the trend of that particular network and how many clients are impacted, what are the applications impacted the network devices?

But if you go into this, and this is where, uh, to answer your question, we have like clients, network devices, infrastructure, connectivity and application. So this entire section is powered by thousand eyes. And here, what I wanna draw your attention to is like a couple of things.

One is the score that we have, like I said we'll be, we are improving our period of time. We have this feedback loop as well. So we not only like, you know, taking industry standards, but we are taking, you know, the client's perspective of like, does this score make sense to you?

Like, are we giving you the right estimation of how your health of your network is? And the other thing more importantly is the metrics that we are looking. So every score is determined by the metrics and the amount of clients that are connected to that particular network.

So we are here for wireless, we have association authentication and IP address for wired, we have authentication IP address and P failures. And again, you know, unlike the other, other sections where network devices see for access points we have AP reboot and low power mode for WAN appliance, we have device utilization. So these are metrics that our customers have been asking, but now we are embedding that in a very contextual way of like, okay, we, if you want raw metrics, that's also available, but if you want it in a way that how it's affecting your health of your network, we are bringing that into on, on this, uh, we are bringing that into the picture as well.

So very quickly, uh, to answer your question, if you have Office 365, which is, you know, this information is being populated by thousand eyes, we tell you like what the loss and latency is over the internet, uh, and we tell you there's a packet loss. So clicking on this packet loss, you can say investigate into thousand eyes and it directly launches in 2000 eyes as well. So that's kind of like the interplay between the two interfaces.

Um, and we are, you know, we are trying to bring that data in and bring that uniformity across the platform experiences. So this is like where you start with the org and you drill down to the network. Now, now that we've, um, again, before I move into the next feature, the score calculation over here, and again, this is again just gonna get better over time, is we are doing proportional weighted average.

That means we are looking at the metrics, we're also looking at the number of clients that are connected, number of wireless clients connected to the, you know, base of the total number of clients that are connected as well. So this keeps changing, right? So the score also keeps changing dynamically.

So you might sometimes have more wireless client than wired or remote, and then if that changes, the score also changes. So we're not looking at it a very, like an average perspective, we are looking at it a very proportional perspective that keeps changing over like, you know, the, the environment of the network. So if you'll only have one VPN client, it's not gonna tank your network score when its performance goes to hell, No, because it's very proportional to like the number of clients that you have have.

So we, we actually, that's the rea one of the reasons where we're like proportionality makes more sense, but like we are improving again, we are changing, we're seeing what gets us closer to the accuracy depending upon the feedback coming in. So now that we've spoken about like, you know, top down, so this is like, hey, give me a view of orgwide, gimme a view of like the health of all the, uh, you know, networks on the org side. Now let's look at from the other way, like bottoms up, right?

You have one client, for example, when you have like 15,000 clients. So like, you know, more number of clients you have say, uh, an executive meeting that is happening and you wanna make sure that these certain number of clients are having a good performance, right? Or one of the presenter is having an issue with WebEx.

How do you really troubleshoot that? So one of the things that you prior had to do is like really understand where the client is, which access point is connected, go to the access point page, go to the switch page, get all that information. You don't have to do all any of that.

We really bring all of this information in a very contextual manner, which already tells you like how the client is connected, but also tells you where it's connected, what is the path that it's taking in the traffic for it to get out to the internet and what is the issue and how can you resolve it. So it kind of answers everything in one single page. So again, you can do this in real time and you can do this historically as well.

So in real time you'll know like how the client is connected, but say like yesterday, you wanna know what the client path was. Was it different? Was it connected to another access point compared to like this access point?

I just had a customer who said like, I was moving from one floor to another and I was getting, you know, complain that, you know, one floor works well, but the other floor wa wasn't. And obviously the, the floor had different access points. So it was hard for them to even say like, which access point had a problem?

But using this was something that was exciting for them because it would kind of like pinpoint them to the right, uh, data, the right information. So Are are these different workflows? You, you have one that's one that's like network score and then you sort of dive down and find the problems and one is I'm helping Bob try to fix a problem.

Are these correct workflows or are they the same? They're The same. So in this example, if you say like, you know, you have like, let's go into this, uh, workflow, right?

Where you had like the client over here that had a problem. So, um, say we were in San Francisco and data center and then there's no client issues here unfortunately, so we can probably Good darn. Yeah.

Uh, San F Cisco. Yeah, there you go. So we have like 212 clients.

So with it, the clients are impacted. So this clicking on this will take you directly into the client visibility. So if you click, that'll Give you the path and all of the other, All of the other stuff, yes.

Okay. Yeah. So that will show you exactly like what the path is and which again, actually let me, let me show you that right now.

So like over here, so say if this is like the front register had the problem, so you go into, so clicking on that will get you into this, which will probably tell you like the entire client connectivity, right? So it tells you like this front register has been constantly running into problems. So it tells you like over a period of last day it's running into problems.

And more importantly, it tells you at this current state it is front register is connected, this AP to this switch to that mx. Now if you go back in time, this will change. So if it's connected to many switches, many access points, we'll be able to bring that as well.

So historically you'll be able to like kind of bring that into perspective too. And Sorry, as we're looking at this, it looks like that this particular register has dropped a number of times. Would you be able to just go back and click the, the times it's dropped and pull up this information and see how fast can you find out if it's wavering back and forth?

So pretty quickly you go, gotta lay off. Yes. Like as you was speaking you saw that I was like changing and it was loading.

So, and the other thing that I want to bring to your, uh, attention is we not just give you like what the client is, but we give you what the problem is. We've set like the failed connection to a society, to this access point is because the DSP server is not responding. So if you click on this, we tell you it's because the DSP lease is unavailable, so you can now go into the respective page and fix this issue, right?

So we're not telling you like how it's connected not just now, but previously we tell you what, what is the issue that it ran into as well and how can you resolve this. What we wanna do eventually is like add tools such as like packet captures over here, which is, you know, intelligent packet captures another capability. So we're gonna be adding that to kind of like give you analysis.

So if you want to be a little bit more like, hey, I wanna get like packet level analysis, we could be able to do that as well. So, um, having said that, uh, the other thing that I wanna quickly highlight is obviously with uh, you know, root cause analysis, there are frameworks that are built which kind of like talks about, you know, which goes into the depth of all the knowledge base articles and all the best practices and the configuration and tells you exactly how the, you know, what the causation is and how you can resolve it. So that is the root cause analysis.

So we've built frameworks that goes across your network devices and your configurations and tells you exactly what you need to do. That's exact, that that framework is what we are using to empower like the causation and the resolution as well across these platforms, right? So we are embedding that into the framework into, into the respective platforms and experiences.

And the other one is alert profile. Now all lot of our customers have different SLAs, have different thresholds. So they will be able to kind of like, you know, configure these thresholds that kind of matches their organization SLAs and get alerted.

So they're not being like, you know, there's no like alert fatigue happening. So do your alert profiles also, um, manage, um, the alerts that come from your MT sensors and whatnot is, is it one framework for all Meraki devices? Yes, One framework for all Meraki devices Including like vape alerts and like all the, all the, you know, high temperature alerts and power alerts, all of that stuff.

All of that. So right now we obviously have like the networking stack. We'll obviously get the IOT stack in as well very quickly.

Okay. So it's, it's okay. It's, it's it's yet to come but it's this, the framework can be used for that as well.

So it's like it's yet to come. Okay. So, uh, very quickly what I wanna end with is like, um, we have a lot of the intelligence baked into the platform.

What we want to do is we want to give our customers a way to kind of interact with this intelligent tool as well, which is our AI assistant. Now what you saw, like we could use the dashboard and you saw how quickly you could also resolve and get to the resolution state and the remediation state too, or sorry, the uh, identification state and the remediation state, the AI assistant will make it even faster for you. So quickly wanna show like a video.

Like if you saw this was where we were at, where it showed you like the problem. Now if the same experience was something that you had to use AI assistant, you could quickly ask, Hey, can you troubleshoot my, uh, front register, which is having a problem? What happened to this client an hour ago?

It pulls up all the issues, it tells you DSCP authentication, you can click on troubleshoot on one of these problems. It tell tells you exactly like the DCP lease is not available. Then you can click on view suggested actions and then it brings up like the list of uh, next steps that you can take.

But if there's like targeted next steps that we have added to our RCA framework, it brings that up as well. So really within few seconds, no matter which page you are in, you can now interact with the AI AI tool and all of these workflows that we've baked in will be a part of the AI assistant tool as well. So that is something that we've made available, um, and a, a assistant is again, a slow roll rollout as well that will be available to all our customers, um, very soon.

So with that, I think I'm up for my time and I really appreciate all of you listening to me and hopefully I'll have enjoyed the session. Thank you.