cPacket Introduction with Mark Grodzinsky

My name's Mark Rodinsky. I am the Chief Product and Marketing Officer at C Packett. I'll be joined in here just a minute by Eric Rudin, our new field, CTO, and Ron Novo will also be joining us.

Um, he presented at both the Security and Network Field Day with me. So I think you remember him. He's the our CTO.

Now, you thought I was kidding when I said, um, you know, the, I was just recently at, at two shows, right? So I was at The Killers if you went to Cisco Live, which was, which was pretty awesome. And then, and then, uh, Metallica ki Eric and I went to, um, here in Santa Clara, which was also awesome.

And there were two things that occurred to me. One was clearly I missed my calling. That is definitely the life.

And the other thing is, it's the epitome of hype, right? You, the, you know, James Hetfield gets up on stage, he yells something and the crowd goes bonkers, right? And it got us thinking about kind of what we've been seeing in the industry in not only our industry, but every industry.

And it's really AI crowd goes wild, right? And then it's, thank you, goodnight, right? Walk off stage.

It's, and but what, what, what is it? Right? It's, um, it's kind of at its peak of hype right now, but in our view, it's a tool.

It's not the answer. And so what we're gonna spend a lot of time here talking about today is how AI without a business and technology strategy wrapped around it, is really, you know, again, it's vaporware. So that's what we're gonna be talking about, how we use AI through our network observability platform to drive business outcomes.

Now, some of you, this is the first time seeing us, so I thought I would do the impossible. We, again, went to network field day 35, we went to security field Day 13, we did 180 minutes of deep technical explanation. And I'm gonna basically give you the five minute, you know, binge watch preview before we get started, because, you know, like every other binge watch show, you watch it for a weekend, you see all 10 episodes, and now you have to wait a year before you see the next thing.

So, okay, well, I don't even remember what happened. So let me catch you up. First, we talked about CPAC and how we've been around since 2007.

We kind of made a name for ourselves at the 2012 Olympics, where it was one of the first 10 gig networks that kind of put CPAC on the map. Um, and we built over time an entire network observability portfolio from packet captures devices all the way through the AI ops. And at the, at our heart, we're packet heads.

We are, you know, we believe packet is the single source of truth melt the metrics, events, logs and traces are absolutely great for, for some purposes. But if you really wanna understand the four Ws of observability, the what, where, when, and why things are happening in your network, you need access to the packets and you need to see it. So, and then we talked about, um, our packet broker, which was, you know, our, our packet delivery architecture where we have an FPGA on every port, which ultimately gives you the, the highest precision and performance, um, for your observability solutions.

And then we went and talked about, um, a very technical explanation about microbursts, right? How is it that ultimately all of our, my network clients are running below the capacity of the network, yet I'm still dropping packets. How could that happen?

Well, if all the network nodes are all hitting you in a very short period of time, you end up with a situation like this kid where there's just packets everywhere, right? If everything happens in a very short period of time, clearly the kid has the capacity to drink that milk. But as you could see, very technical explanation of microbursts.

Then we talked about our reimagined packet capture and analytics up to 200 gigabits per second, right? To disc concurrently with indexing. So you can very quickly recover your packets, or sorry, your, your captures and analytics.

So all of that happening at line rate 200 gigabits per second with line of sight to 400 as we move forward. And finally, we wrapped up and we were accused of burying the lead a little bit, but we wrapped up with an introduction to our observability AI solution. It was, uh, kind of the companion NetOps type of a solution where, and that was where we introduced it.

And then we ended our first network field day. Then we went and, you know, bravely stepped into unfamiliar territory, coming into a security field day, right? Again, we're, you know, network people at heart.

But we strongly believe with the net, the knock stock convergence that's happening, it's more than just operational efficiency that's bringing these things together. But having your network team and your security team working from that single source of data, that single source of truth for both of their solutions is ultimately incredibly valuable. And the right networking solution and op, you know, observability solution will augment and enhance your security posture.

And we went through a whole presentation about that going from how we can aid in protection and how we can detect more efficiently, how we can respond and do digital forensics, and how we can validate and show compliance, right? So we went through sections on each one of those. We showed demos for each one of those.

So again, if you have interest in seeing that, again, I would encourage you to, to go to the YouTube channel and, and check it out. And then we figured, you know, we showed how AI could also help, right? And if you don't have a deterministic threshold, but rather you wanna just monitor the network and see how it's doing, then AI can help set those thresholds for you and it can actually, um, be a smart companion on the security side as well.

And so that's that kind of wrapped up our network field day and our security field day. So that brings us to why this all matters, right? And what we're gonna talk about today is how you need, what really this, the what CP a's all about and what network observability is about is service assurance.

And we just had another great example, uh, of that just a few minutes ago, right? Observability to proactively assure service reliability, because obviously it needs to be pervasive. We need to see it everywhere.

We need to see every node 24 7. It needs to be independent and open for integration, right? So we don't hoard our data to ourselves.

But rather if you've already got a big Datadog deployment or you have a big ServiceNow deployment so that you, for service doing your tickets, the information that we have has integrations where we can then feed that to all of the different tools that you've got and all open APIs and Eric will go into the architecture of how we do that, and ultimately scalable, right? Because we're talking about trillions of packets that are going over these networks every day. Tens of billions of sessions, right?

But there's thousands of dashboards you could be looking at, or can we distill that down into these handful of insights that will allow us to act and understand things now of, you know, and, and do that proactively so we can prevent some outages from happening. So the rapid inaccurate detection and prevention of future failures, that's really what, you know, we're, we're trying to do here. So with that, um, this is the foundation for the rest of the presentations.

We're gonna show how AI and the overall observability architecture can be used in different environments and really show how we can drive business value. Eric, Thanks Mark. Take it away.

Awesome. Thank you all. So good morning.

Uh, thanks for coming out and for those online, thanks for stay sticking with us as we go through our little blip there. But we're back. So, you know, I've been with, um, CPAC now for about three months and, you know, got an opportunity as field CTO to meet with a lot of our customers.

Um, one of the things that I hear a lot more is cloud has made networking cool. Again, AI is influencing that. Um, and that is because all of our customers run either hybrid or multi-cloud, right?

They're not just all data center, they're not just all colo. Um, they're not, some are all cloud, but obviously there's gonna be some connectivity to different services if they're distributed. But there's a commonality of a lot of our customers, and that is they have mission critical applications that have to work if it's financial trading, if it's healthcare, if it's government services.

And this is just a, a basic representation of the kind of complexity that a lot of these enterprises and large organizations have to deal with. There's different connectivity points. There's SD wan, there's firewalls, there's complex routing, there's layered switches, there's top of rec switches that connect through fabrics, right?

And so it's essential, as we think about service assurance and we think about how to get organization, we have to start looking at a way of tracking that. And that is where monitoring points on key links is essential at your observability strategy at the core, which is that is our, that's our point of getting data through either span or tap ports that we are able to aggregate and start collecting data and then do some of the deduplication allow for the load balancing and start moving this data into a central location that we then do analytics and ultimately share through our o open APIs. And that is a core strategy of, of c packett to provide that pervasive hybrid cloud, multi-cloud, multi-data center, uh, view into the entire infrastructure at the essential points.

That is really the key key, uh, element. So as we, um, look through our solution, it starts with that data we have to tap and get and get those packets coming into the system. And that's where we start with packet acquisition with the smart ports.

And this is where we start doing timestamping. So one of the unique differentiators of our solution is that we're doing timestamping in nanoseconds. And so we're syncing, syncing that time through different protocols.

We're aggregating that data, um, over our switch fabric, but we're also collecting, uh, metrics at the time of acquisition. That is a key differentiator. Think of it as like a probe inside the switch.

And so by doing that, we're now starting to build the observability logic of the cus customer's traffic characteristics. We're able to start identifying patterns like microbursts. We're starting to see that latency at a very low level.

And so when you get into services, and one of the things we're gonna show you today is very specific examples and very specific customer use cases where this is absolutely essential. We also collecting that, that shore, and you'll see where that comes very useful as we look at this holistically as a platform play. We also integrate those packets into the ecosystem.

It starts with our own solution on our capture devices, as Mark talked about. And this is where we start building session analytics. This is where we start correlating different types of data, like what was the source, what was the, the host name, how do we get aligned, the DNS or LDAP or HTTP metrics.

We start doing TCP analytics within the system. We're able to build topologies. We're starting to look at more sophisticated use cases by taking that stream.

And then building, building the analytics in a, in an enhanced and enhanced layer inside the capture technology. We also connect that to external storages for larger environments that have high compliance requirements where they have to store every packet. They have to be, they're concerned about things like exfiltration.

They want to understand lateral movement, they need to know where within the system the data is moving. And all of that is wrapped into our analytics visualization layer. And so we have a metric store at the top with through consult center, that's the management, uh, foundation.

And then we've got on the right different uni, uh, visualizations powered by Grafana. And then we will start introducing this concept around M-C-P-M-C-P if you guys aren't familiar with it. And then we'll talk more in one of the use cases is around building context.

It's model, uh, model context protocol. It's, think of it as like a UB, uh, USBC of interconnecting different devices. But this is the foundation for how different AI acentric workflows talk to each other.

And we're gonna go into more detail later in the presentation. And then Mark also talked about being open. This is one of the essential elements is that we don't hold our data back.

We actually expose it at every different step between all the different elements of our solution, including the cloud components. Everything can be integrated into your existing tool stack, and that's part of the use cases that we'll talk through. So I think this is really important to think about innovation and think about how AI is useful, but not, it's not the, the, the core of what we do.

It's an enhancement, it's an augmentation of a lot of the existing analytics, but it allows us to drive more business value by providing better visibility into the anomalies so that all of these billions of sessions that are happening can roll up to an anomaly. And it's really important for that because when we start thinking about putting the AI to use, we think about, uh, uh, unutilized links, overutilized segments that might relate to, uh, some sort of degradation. So we can be predictive.

This also could relate to security issues, and we're gonna start showing those later in the, in the different use cases. But then we know that as, uh, these LMS mature, this is the new interface, this is the new query, this is the new dashboard, LLMs, they're different flavors. We're also allowing you to bring your own.

And so this is really important. If you think about different industries and industries that have their own private models, industries that wanna run all of this stuff behind their firewall. They don't need to have it, but we can embed their, whatever interface they have into our architecture to allow them to ask the information and get the data quickly.

And then as well, uh, we've talked, we we're gonna not get into too detail into the MCP but model, but I think this is where in the future, ag agentic workflows are gonna allow systems to talk to systems. And that's all gonna be based off of a common protocol. And there's gonna be, there's different ones that we're introducing, but we've been investing in that.

And that really helps with the ecosystem integration. It helps with these different types of workflows that we're building with our customers. So we're gonna introduce today this concept called the, the value equation.

We're not here just to talk about packets, right? Packets are useful, packets are informative, packets provide the source of truth. But really what, what this is, is about outcomes.

When we talk to our customers, we're not saying the answer is just, you know, you need to catch this P cap at this time. It's more what problem is really driving you nuts? Are you running a streaming service and something happens on the core internet and now that streaming service doesn't work as an example, what do we gonna do?

We're gonna take that data, we're gonna enrich that data with other context. Context comes from different systems, context comes from other services, applications, um, different tools. We can learn a lot about that.

We can also start to build metrics and, and observability data at that time of the enrichment. That's where we're adding more information. But that doesn't really get you there unless you're able to then enable the AI power that we now have access to, to really get that insight.

But that insight, staying sitting inside of a, of a, of a small data set is not really useful to the customer until it gets integrated into how they do work. So we've been calling those, uh, business workflows and we're gonna walk you through this framework across all of our different use cases. And that workflow also is where the ecosystem, either if they're receiving our, uh, packet feeds or they're receiving our data that we're streaming out of our analytics layer, uh, out of our control center, they can, we can, uh, send that data out to their data lake.

They can do their own analytics. This is all in real time so they can solve that problem faster. And that's where we get measurable outcomes.

And so really what we're gonna challenge ourselves today through each one of these different use cases is how does this impact the customer? How does this network observability element help solve the problem? And what, how can AI help us?

And so the way the, the guide for us for, and you'll see this applied to all of our different use cases, follows a basic simple customer journey. So this is all about the customer. How can they consume this technology?

How can they start integrating into how they operate? And the first thing is, and this is why your, your role is absolutely essential in this discussion, which is understanding the, the bus, the customer's core business, understanding the things that are driving them nuts, understanding the current network conditions, right? And this goes back to their path to 400 to 800, what are they doing around measurement?

And then how do we integrate this, this technology into all of their existing tools, but then also show them that we're able to get that data at the at line rate, get it into the system that allows them to contextualize it, add the AI insights, and then get that connected into their whole tech stack. And if we do that well, then we have the validation, right? And the val, the validation, sorry, is only useful if we can do it again, and we can do it again and do it again.

So this is optimization. And then this is continuous improvement. So if we do our jobs well as network operators, as security operators, we've in, we've injected technology, we've injected packet based data, we've injected observability metrics into how they run their business.