NB565: New Algorithm Claims Quicker Quantum Decryption; NVIDIA Revenue Rockets Higher

0:00Take a network break. I'm Drew Conry Murray. John and Johnson and John Berker are ways from pleased to welcome Tom Hollingsworth from the Futurum Group and Tech Field Day as guest opinionator. Tom, nice to have you again. It's always a pleasure to be here with you, Drew. Well, thanks. We got a bunch of stories to cover, but first day reminder that after the news, we have a sponsored tech bites conversation with StatSeeker, they're a network monitoring company that collects high fidelity network data to help engineers and administrator get visibility into physical virtual and logical interfaces to find and fix problems. We're going to explore how StatSeeker works here, customer use cases and find out how StatSeeker differentiates itself from other network monitoring products. And by the way, if you like network break, we've got a whole bunch of other podcasts over at packapushers.net, including day-to-cloud, pack-a-protector, heavy networking, IPv6 buzz network automation nerds, and N is for networking. It's a nerdy tech analysis and compelling conversations all about infrastructure cloud, professional development, and more. All right, we'll start off with our red alerts. Cisco has issued two CVEs for its Secure Firewall Management Center or FMC software, each of which has a risk score of 10 out of 10,

1:02great for the Olympics Gating Terrible for CVEs. Both vulnerabilities are tied to the web management interface for FMC. The first could quote, allow an unauthenticated remote attacker to bypass authentication and execute script files on an effective device to obtain root access to the underlying operating system and a quote. The second could also allow an unauthenticated remote attacker to execute arbitrary Java code as root on the effective device. Regarding the second vulnerability, Cisco notes the attack surface is, quote unquote, reduced if the FMC management interface doesn't have public internet access. However, Cisco says there are no workarounds for either vulnerability. Cisco has release software updates. You have to apply these software updates if you want to fix the bugs. Tom, I feel like calling anything secure when it's related to security is just like asking for it. Yeah, it's only a matter of time before somebody finds a hole. That's one of the things we teach a lot of people when they're starting out in security, right? Is there's no such thing as perfectly secure or perfectly patched? And there are ways to alleviate some of those issues.

2:03Like one of the things that Cisco said, why on earth are you allowing access to FMC from anywhere that's not internal to your network? Like, you know, we think all the way back to like 1996 mission impossible, right, with the black vault that William Donnelly was in. He may have been onto something there. It's like one guy with access to one server to be able to fix things. Yeah, it's a bottleneck, but I promise you, unless your Tom Cruise, you're not hacking that thing. I guess I'm feeling like that's not really scalable. So this is, but this is the world we're in. The problem though is that people are scaling their ability to find these kinds of vulnerabilities in their attack surface. And if you're not scaling along with them, you're going to have to scale way back. Yeah, that is also true. All right, let's move into the news. Sassy provider Kato Networks has announced a new offering that to my mind kind of sounds like automated threat hunting, although that's not what Kato is calling it. They're calling it dynamic prevention. The feature collects and correlates anomalous signals over time

3:03to look for patterns that might add up to malicious behavior. If the system determines these signals, do indicate some kind of malicious action. It can automatically apply rules to block activities. Kato says the service, which is included in its Sassy platform can correlate months of network and security activity from Kato sensors, including DLP, IPS, and anti-malware. This is an interesting idea. I think it, and it does kind of fill in a gap where, you know, you just sort of point in time, security controls might not catch an individual action or an individual signal, but they are part of a longer attack chain. And if you can't afford a sock or you don't have threat hunters in your network, this might be worth a look what are your thoughts? I think it is because we're starting to see this more and more with the rise of using automated tooling to do these kinds of attacks. You're in essence kind of fuzzing the network to try to figure out where you can get entry points or what can slip past the sensors. And if you can gain a foothold, then you can deliver some really nasty payloads. And I think that more and more companies are going to start doing this as they gain access to using AI tools,

4:04be able to correlate this data more effectively. Because one of the things that we've noticed over the years is that while humans believe they're really good at pattern recognition, they actually really suck at it. And so having a system to kind of sit there and say, wait a minute, we're seeing, you know, we're connection requests about the same time every day from roughly the same area of the world that are trying to deliver these certain kinds of payloads. You might want to look into that. That's just one example of being able to put those pieces together to kind of raise suspicion. And yeah, it really is, it kind of looks a lot like threat hunting to me, but people don't buy threat hunting. They buy dynamic things or they buy preventative things. Because prevention is an op-ex cost, not a capex cost. So people are more than willing to pay monthly for that. Yeah, threat hunting is definitely a capex cost. I think the key thing here is like the time element of months. Because as you said, sometimes humans aren't great at pattern recognition, but if that pattern is happening in a day or in a few hours,

5:06maybe you would pick up on it, but if it's happening over weeks at a time in a stealthy attack, that might not make it into your eyeball. So it's an interesting idea. All right, we've got a few quantum stories to cover. First, Google has announced how it plans to support quantum-resistant cryptography in its Chrome browser, rather than add X.509 certs with post-quantum cryptography to its Chrome root store, which Google says could introduce performance and bandwidth issues for TLS connections. Google's going to rely on something called Merkley Tree Certificates. I don't know if it's Merkley or Merkle. Anyway, Merkley Tree Certificates or MTCs. MTCs are part of an IETF working group looking to address the impact of larger post-quantum signatures on public key infrastructure. So what are MTCs? The Google blog says, quote, MTCs replaced the heavy serialized chain of signatures found in traditional PKI with compact Merkley Tree proofs. In this model, a CA signs a single tree head representing potentially millions of certificates.

6:06And the certificate, this air quotes that Google uses, sent to the browser is merely a lightweight proof of inclusion in that tree. Google says MTCs will make it easier to drop post-quantum algorithms while avoiding bandwidth and performance penalties of classical X.509 cert chains. Thoughts your Tom. So I see this being proposed a lot in IOT devices when I talked to Digestert Mike Nelson years ago was talking about the fact that if you try to throw a big RSA 2048 certificate at your thermostat, it's going to barf, right? Because your thermostat CPU is not designed to handle that kind of calculation load. So I hear what Google's saying about bandwidth and performance for TLS, but I can't help but wonder is this an effort to make those low-cost CPU ARM CPUs that are going into Chromebooks last a little bit longer? Because as everyone knows, like PQC, one of the reasons why it works as well as it does is because it's doing lattice-based encryption, which means you have a lot of calculations that have to happen, which makes it really hard for classical

7:07and quantum computers to break it. But that does increase the overhead. Honestly, with today's internet, I don't think that the customers are going to bat an eye with an extra lost second or two on a connection. Yeah, some people are. But realistically, this is about devices that they want to put Chrome on that are not going to be tolerant of the extra processing load. Now, I think that that will change. I really see this as a stop-gap measure for the next few years to start seeding values about store-and-harvest attacks where people are collecting data that exists now and then they're going to replay it later against quantum algorithms that are optimized for cracking things like RSA and PKI. So I think that what you'll see is probably within two to three years when better ARM processors come out when more robust CPUs come out, we'll see this start to drift away because those processors will be optimized to do certificate generation and key exchanges and things like that. Kind of like we saw with Intel where they had cores that were dedicated to doing that.

8:09We'll see that kind of start happening as the software becomes better optimized for this because it is the reality that we face now. This is no longer just somebody's idea of what's going to happen if someone knows how to break RSA. We know what's going to happen, which is something we'll talk about in a second. Yeah. Yeah, that's an interesting point about Chrome on ARM processors and IoT devices. I'm sure Google wants Chrome to be everywhere. So it needs to be able to one run and support PQE without falling over and to make Chrome viable everywhere. So yeah, I can see that. Just to wrap up on this, Google is adopting a multi-phase trial and rollout phase one, which is currently underway as testing the feasibility of MTCs by using MTCs with real internet traffic, but they're backing them up with traditional X.509 additional phases will begin the following year. We've got a link to the Google security blog if you want to get more details. I was sticking with quantum Canadian researchers have published a paper claiming to have developed an alternative to shores algorithm that would require substantially fewer qubits

9:11to crack RSA 2048 from approximately 1 million qubits using shores algorithm to the low thousands. They also estimate that theoretically the new algorithm could take just minutes to crack RSA 2048 versus years for shores algorithm. Their method blends the use of both classical and quantum computers to achieve its results. And I just want to say at the outset here, these researchers are not claiming they have cracked RSA 2048. They're saying their algorithm based on the tests that they've done could potentially someday be faster at it than shores. So leave it to Drew to give me something that I actually had to do research on. This was kind of fascinating. This paper was cool. Basically what they're doing is they're saying shores algorithm relies on a furrier transform, right? So there's a lot of compute power that has to go into that. And they're essentially saying that they found a shortcut because what they're doing is they are looking for instead of doing processing over a series of complex numbers, they are actually looking

10:14in finite fields for modular arithmetic. So basically they're saying we're seeing patterns and we're kind of making sure like we're working on those. And the other thing that people have to understand is there are huge quantum computers out there. Like every day they're just getting more and more powerful. And that's what the qubit number is. But the problem is it's not just the number of qubits that the computer is capable of producing. It's the amount of noise that comes out of it because there's a lot of videos out there you can watch on it. But basically what happens is when you're trying to find the state of a single qubit, all of the other states are noise that consume other qubits. And through the magic of math, whatever you're looking for actually ends up being the right answer if you know which qubit to look at. You've just got to filter all of the noise. And that's why we haven't cracked RSA yet is because the amount of noise that's generated as you continually add qubits increases, not exponentially, but it increases in a similar fashion. And so we have to get better at developing software

11:15to filter that noise. What they're doing here with this particular algorithm is basically saying, yeah, but what if we didn't have to worry about the noise because we found a shortcut in the math itself? And it may not sound like much, right? I think one of the things that quoted in the paper was there was like a, in theory, there was like a 33% reduction. But in practice, it was closer like 40%. And you're like, oh, well, that doesn't sound like a whole lot. Yeah, well, theoretically, in order to crack RSA, you need effectively a million qubit computer. If you can reduce that to 600,000 qubits, that's a pretty big reduction. And when you consider the amount of power and things like liquid nitrogen that it takes to do this, because yes, we have to keep these things stupidly cold right now. Well, some of that depends on the type of quantum computer, but yes. And that's true. And that's one of the reasons why we're starting to see this approach with a hybrid classical and quantum computing is to allow the quantum computer to do things that's really good at and not have to consume resources to do everything. And I think that that's where we're going to see the biggest breakthroughs to start with is quantum computers are going to be kind of used almost like a deep thinking core to do these kinds of things,

12:18to instantly solve these factorization problems. But then it's going to shovel that data back to a classical computing core, like a cluster to do the actual work of implementing it. Because otherwise, I mean, we're already seeing this right now with AI data centers, right? Like, we can't get enough resources to run these things the way that people want them to be run. And that's a quote unquote new problem. Quantum computing, we've known about this for decades because we know what it takes to run a quantum computer. And even with the advances that we've been making in the science, it's still really hard. Yeah. Yeah, their method did blend the use of both classical and quantum computers to achieve their results, which I think also happens when you're using Shore's algorithm to attack cryptography. The researchers say they have tested this algorithm. They call it JVG, which is the first initial of each of the researchers' first names. They did a simulation and they actually did use a quantum hardware from IBM and they're testing. Again, they didn't test it against RSA 2048, but they did test it. And they, at the paper notes quote,

13:18on quantum hardware, JVG reduced the required runtime from 67.8 seconds to two seconds and the quantum gate counts by over 98%. I should note that the paper has been released as a preprint, which means the version we've linked to hasn't gone through peer review or been published in a journal. That is not an indicator one way or the other of the paper's quality. I think preprints are a standard practice in academia. I just want to make sure that listeners had that context. And we were talking about quantum computers with the number of qubits I was curious what's out there. Based on the quick research I did, Caltech is currently the winner. They rolled out a quantum computer last September, which supports 6,100 qubits. All right, moving on, quick update from NVIDIA. The company recently pledged to invest $100 billion in OpenAI and how NVIDIA is walking that back. It's capping its investment at $30 billion. CEO Jensen Huang said it was because OpenAI is going public soon, not for any other reason you might think.

14:19There has been criticism of sort of like these circular deals by NVIDIA and cloud giants. They promised to invest billions in the AI companies, and then the AI companies promised to buy billions in products and services, which has some people in the market being like, wait a minute. So Tom, does this feel like a walk back or a reality? Is this a crack finally in the great infrastructure boom or? I don't know if it's a crack, but I think it's maybe admission that just pouring money into these things isn't going to be the end all. Jensen's got a good point. If they're going public, why do they need my money now? Why can't they just buy things from me like normal companies do? Because you bring up a really good point. If I'm investing $100 billion in you for you to turn right around and buy $100 billion with chips back to me, then effectively it's a net zero. Although to those people who pointed out, this is not in-ron because this isn't fake money. It's actually real money. It's just, it's on a dump truck being driven back and forth between open AI and NVIDIA all day long.

15:20But the problem is that it's still a net zero. And so I think Jensen's point is, if you are about to go convince a whole bunch of investors to invest in your company and give you lots of billions of dollars, maybe it's time you give the actual billions of dollars to me for the hardware that you need to make this work. And I think that he realizes that if he doesn't scale back now, then they're just going to create this feedback loop. And this is in a way, it's like weaning a calf, right? Like I've got to give you a little less to see if you can make it. And if you do, to extend the analogy, I'm fattening you up for the slaughter later. Because, you know, well, with all the announcements that we've seen this week with open AI making deals with the federal government, I'm sure Jensen's like, hey, the government's got money. I'll take theirs. It spends just the same as yours. Yes, for sure. All right, moving on this week, Accenture. Greed to Choir. Ucola from ZIF Davis. Ucola owns the popular speed test website, as well as wireless design company. Echahau was sold for approximately 1.2 billion.

16:21The organization's going to be folded into the connectivity division of the consulting firm Accenture announced that they will target Ucola services for end-to-end network intelligence services for AI-based transformations. Tom, you brought this story. And I didn't think of Accenture as like a product and services company. I thought it was more of a consulting company. So I think this is odd. But maybe I need to update my mental model of Accenture. So it's funny because this actually came out earlier this week. And I was talking to some of my Wi-Fi friends, like Keith Parsons, host of heavy wireless. And we were debating this back and forth. And one of the things that came up was, why did they pay so much money for this company? And one of the things was, when you think about what they are getting from Accahau, which is a component of Ucola, which is now part of them, they have a lot of data on Wi-Fi installs. And there's some questions around, well, who actually owns that? And I think there was some clarifications in the last couple of years in their user agreements of exactly who owns the data and all that other stuff that fits up with it to the cloud. You know, that's going to be something valuable for Accenture because what they're going to be able to see is the quality of Wi-Fi deployments out there.

17:23But the flip side of this, and this is something that I was thinking through as people were bringing up another product that Ucola has come out with in the last year. It's an assurance system for hotels or other large venues. So basically what it is, you go in and Ucola will sell you a service that will allow you to attest that the Wi-Fi in the hotel is good. Like, you know, three, four, five stars kind of thing. Right. But when you think about it, like when people go shopping for a hotel now, yeah, I'm looking for a spa, I'm looking for a pool, but if you're a tech person, I'm looking for good Wi-Fi. I'm looking for good connectivity. And what better way to do that than have a third party offer that as well, you know, once a year we'll test it to make sure it's working or in Ucola's case. What they would probably do is they would put a speed test server on-site and they would use it to constantly monitor that system because it doesn't cost them any extra. And then they can update that in real time, right? What better way to get into a consulting company, like Accenture, than to provide a service that they can add on to your yearly contract

18:25and basically make all the money back that they had for doing that. But now that I have all of this data about the Wi-Fi installs, now I can go back and sell you a service and say, hey, you know what? I can make that connectivity even better for you so you can go from three to four or four to five, which is something that I can prove over the time that I have all of this extra data about how many more visitors you get when your Wi-Fi is better. And that will not only offset the cost of the upgrade, which we will happily sell you, but the cost of the service to keep it maintained all year long. And someone in Accenture is probably like big dollar signs flashing in their eyes because of how great of an idea this is. Yeah, 1.2 billion is a surprising number for this portfolio, again, going to Accenture. But I guess I can see the argument for it. All right, we'll finish up with some financial results. We'll start with Broadcom. They announced results for their Q1, 2026. They are definitely writing the AI infrastructure wave. Revenues were a record 19.3 billion, up 29% versus this time last year with net income of 7.3 billion, up 34%.

19:28The company specifically called out its AI-related semiconductor revenue, which was 8.4 billion, up more than 100% since last year. Broadcom software business, which includes VMware and Symantec earned 6.8 billion, which is a nice number, but it was only up 1% versus this time last year. So you can see where all the energy is. You remember when we thought that hardware was dead and software was the key to making all that money? All we needed to do was create a paradigm shift and all of the sudden, now everybody wants semiconductors and that's what we're seeing everywhere, right? Is hardware is selling as fast as they can get it cut off the die because of the potential of where we're at? Like that's the best kind of hardware sale to these companies because even if you don't actually end up using it, I have sold it to you. And so I think that that's one of the things that we're seeing is that Broadcom really understands that they have a mint sitting underneath them because it's not just the chips that are being bought for GPUs and CPUs. It's the network infrastructure that runs it

20:28and who supplies most of the chips for that network infrastructure? That would be Broadcom, yep. Yep. Yeah, and for the next quarter, the estimates also look great. Broadcom's forecasting 10 billion in AI semiconductor revenue alone for Q2, as well as a new record for their total revenue. So the tide has not dropped here. It is lifting other boats as well. At least those boats in the semiconductor business Marvell reported Q4 and full year fiscal 2026 results for Q4 revenues were 2.2 billion with net income of 396.1 million. For the full year, Marvell had revenues of 8.1 billion, which is a record. Up 42% year over year, they brought in net income of 2.6 billion. Marvell is also forecasting strong revenues for its next quarter. So Marvell also enjoying that. We need to buy all of the chips everywhere mindset. Yeah, I think that this is actually really good news for Marvell. I mean, they are a part of NVIDIA. So they are getting a lot of that kind of halo effect. But historically, Marvell has been seen to do a lot of infiniband work.

21:30And I think that as we're starting to see the shift inside of NVIDIA going over to Spectrum X, I mean, obviously, those are Marvell DPUs and they're using a lot of Marvell semiconductors. Like, that is the thing. If you buy the package from Jensen, it includes all of the stuff that is made by in-house companies. It just makes sense that if people are buying tons of GPUs and tons of servers, they're also going to be buying tons of networking equipment and storage equipment on the back end and make it all work. And why not sell as much as she can before people figure out what's going on? Yeah, exactly. All right, our last start for the day is a crowd strike. They are security vendor. They reported Q4 and fiscal year 2026 results. For Q4, revenue was 1.31 billion, up 23% net income was 38.7 million. For the full year, crowd strike had revenue of 4.81 billion, up 22% however the company lost money for the year, posting a net loss of 162.5 million.

22:30I will note that crowd strikes spent more than a billion dollars on acquisitions just in January of this year. They bought a browser security company and an identity security company. So they, I think, are willing to lose some money in a bid to grow into new markets. I think that what's happening with a lot of these companies that had name recognition is that they are trying to diversify across all these markets in order to be able to pick up additional service offerings because a lot of companies will be willing to say, well, I know crowd strike and I trust them. Relatively trust them after the whole kerfuffle a couple years ago. But I'm going to buy the whole suite from them, right? Because I don't want to do this piecemeal and have to worry about tool integration and things like that. But the flip side of that that we've also seen is that other companies that have a similar offering maybe not necessarily in security are now buying into security companies to provide that one that popped up a few weeks ago with service now. Like they made a play for security companies so that they can start offering security services on top of everything else they do. So I think this market's going to get very crowded

23:31and a lot of the smaller players are going to find themselves getting bought out to be features in a larger service set. I mean, we saw that with Zscaler and SquareX recently where Zscaler purchased SquareX to offer browser security as a part of what they were doing. That's something Drew, you and I talked about on a recent tech field they podcast episode about how that seems to be a new market that people are going to get into. So I wonder what's going to happen with all the other companies that are kind of playing in that market if, you know, company like CrowdStrike are going to say, you know what, it's time for me to pick this up because it's an offering that I can add, I don't know, $2 per seat to the bottom line. Yeah, and browser security in particular is becoming a hot market in the security space in CrowdStrike, you know, sort of specializing in endpoint protection. I assume they feel like the browser is kind of our territory. So we can't, you know, let somebody else with a secure browser get onto a platform that we're also protecting. So it makes sense that they would want to go that direction. All right, links to all the stories with COVID will be in the show notes that accompany this podcast. That does wrap up the new section.

24:31Tom, working folks, find you online. The easiest place to find me is techfilterlay.com. It's all the stuff that we do for all of the fun going on. I also do a couple of other podcasts to techfilterlay.com podcasts, like I mentioned, security bill of our podcast. But of course, if you want to see my ramblings about all the other random stuff that I do, networking.net is my blog. I'm trying to put some more stuff out there for people to enjoy and possibly even get mad at and leave funny comments. Absolutely, it's a great blog. We feature your posts often in our human infrastructure newsletters to go check it out. I'm Drew Conner-Mari. I'm on Blue Sky at Drew C.M. and I'm bloggingapackapprosters.net. Please do stick around for our sponsored techbites conversation with StatSeeker. They are a network intelligence platform and you can learn about what they're up to because that's coming right up. Today on the TechBites podcast, we hear from StatSeeker. This is a network monitoring company that collects high fidelity network data to help engineers that administer to get visibility into physical, virtual and logical interfaces to find problems faster, understand root causes and spot behaviors and anomalies so that you can prevent problems

25:31instead of just reacting to them. We're going to explore how StatSeeker works, here customer use cases, and find out how StatSeeker differentiates itself from other network monitoring products. Our guests are Dylan Hensler, customer solution specialist and Andrew Greenlaw, technical account manager. Dylan and Andrew, welcome to the podcast. And Andrew, we'll start your first. Just give us a quick overview of StatSeeker. Thanks, Drew. So StatSeeker is a self-hosted platform for monitoring critical networks, some performance metrics, and events like you spoke about. It's really around real-time troubleshooting but then also that piece around the historical analysis, long-term reporting that StatSeeker is kind of known for. It's an on-premise solution. So there's a lot of flexibility options for how it can be hosted. We've got customers that still have this in physical and virtual environments, but then have moved into more Azure, AWS, Google Cloud, we've even had customers pushed us into Nutanix and things like that. So it really kind of gives customers the option to place this where they want and kind of control and manage that data themselves.

26:32You know, not bound by a SAS solution. This is their data. So it doesn't need an army of pollers. It doesn't need separate databases that you're managing. Everything comes completely housed inside StatSeeker. So it really is a smart, highly efficient data collector for a range of different network sizes. So we go up to the super large, but we also play in the small and medium-sized networks as well. Well, Andrew, StatSeeker has been around a long time. I mean, we've done work with you as packet pushers going back. Maybe a decade. It was you've been around a long time. But the thing I remember about StatSeeker from back in the day was you guys could ingest a stupid, ridiculous, crazy amount of telemetry and handle it. And like you were saying, without having to have 50 pollers out there to be able to scale, is that still the thing? Yeah, that's still very much kind of who we are and what we do. And you say a decade, but we've been kind of around in this kind of space for past 25 years now. You know, this is our kind of bread and butter network performance monitoring.

27:34We've kind of seen a lot of change in the environment and the kind of the evolution of networks. But definitely that whole value around data, data collection, large, big data play. That's kind of who StatSeeker is and that's very much kind of what we do. Regardless of kind of the change in, you know, be where the data sits and things like that. You know, so, yep, definitely. Let's dive into what kind of data are we talking about. Foundation of what StatSeeker is looking at is just your classic network metric device and interface performance and health stats. We're primarily collecting those over SNMP. And we're also doing reachability and event and latency signals via ICMP. So very open protocols. We tried which will just support almost any network device out there. For discovery, it's all IP range driven and then configuration and inventory details give you refreshed over time through your scheduled re-walk. So the system is going to stay aligned with what's actually on your network automatically. That's a key thing for us.

28:35So you set StatSeeker up and it run a lot of it can be automated in the background and it maintains itself. There's not a lot of day-to-day maintenance or configuration to do. On top of that, StatSeeker can work with event data like CISLOG, SNMP traps for alerting and troubleshooting. It also supports integrations where the data source is a vendor API. CISCO ACI or Marockie integration are good examples of that model. And then if you have specialized devices or metrics, we do have what we call custom data types to let your extend what can be pulled and reported. Okay, so you're bringing in data from everything and you're doing it in a modern way. You can pull in data, like you said, VA API, if that's what I want. You can grab my data that way too. Okay, so specialized integrations then, like you mentioned CISCO ACI and Marockie, does that mean like yeah, we did some work to make sure that I can model that specific sort of network device? 100% we've got both of those fully built out in the product

29:36and we ship with built-in dashboards for those. So you connect StatSeeker to your API key for those and you're up and running within minutes. So there's really not a lot of configuration to do. You can, we're very flexible, but there's not a lot of configuration to do out of the box. But if you really want to get narrowed down to a specific picture, we can do that as well. How often are you collecting this data? Of, or SNMP, we're pulling every minute for the API integrations. We're pulling every five minutes. And just the big thing there is that we store those five minute or one minute pulls indefinitely as long as you've got the space for it on your server. So we don't roll anything up. We don't average it out. So you never lose granularity? You never lose granularity. Yeah, that's a big part of what we do. So you can go back months and see exactly what happened in real time for any of your devices or interfaces. A lot of these platforms, even like Muraki, for example, you know, they're doing a huge amount of heavy lifting around configuration

30:38and there's a monitoring component. But they very much push you to the API and things like that. If you want to start doing kind of more analysis and extrapolation, things like that. And so that kind of integration really kind of does well with how we then collect and store the data. And then you can kind of slice and dice it however you want. So, I mean, typically we think of the network monitoring is on, because we care about our physical devices, those things we've lovingly racked up. But I'm going to have stuff all over the place, things in the cloud and so on. I'm going to guess you can monitor all the things, yeah? Yes. As long as the device has, we have an API integration for it or we can do SNMP. We have some way of talking to that device. Yes, we can. And then we can integrate it with our databases. So you can really put all your data together from different sources. So remote distributed environment, wherever I don't have to think hard about this. Sure. So you can play it on-prem as a VM in the cloud environments. We do have observability appliances for distributed networks.

31:39That appliance can run monitoring services from a remote location. Example services would be ping-pulling, ping-only discovery, things like that. And then we just forward the results back to your central sat-seeker server. And it combines it all together into one view. So you get the reporting and dashboards from various sources distributed around your network. So centralized data store, but I can have observability out it. My distributed edges would be- Yep. And that really lets you get consistent visibility across sites. And then you don't have to redesign your monitoring around lots of separate tools that way. Okay. So once you're gathering this data, what are you doing with it? What am I as a customer looking at? Sure. So it is a huge amount of data just because we're pulling so often and pulling in so much. And then that doesn't really do you any good if you don't have anything to do with it. So there's two big things. It turns data into answers right now and evidence for later. In the moment, all that data is driving dashboards, reports, alerts. So your team can see what's changing

32:39and respond quickly. And then over time, because we've got all that historical data at full resolution that we were talking about, you can replay incidents, do root cause analysis without losing detail to averaging a roll-ups or aggregation. Then there's also the proactive side. Statsheaker's reporting engine uses historical baselines to model typical behavior. So that supports forecasting, trend analysis. And that really allows you to spot bottlenecks or shifts before they turn in outages. Give me an example of a dashboard. How would I construct Statsheaker dashboard? Maybe like a typical customer what they do with a dashboard that you've seen be useful? Sure. Our dashboards are all GUI-based. It's very similar to Grafana if you're familiar with Grafana. So we offer a full library of pre-belt panels for most of the network metrics. So a good example would be utilization on your receive and transmit side. You can select a few interfaces, select the panels we've already got built for you

33:40and drop those onto a dashboard, specify your time frame, what devices you want to look like. Statsheaker will start graphing that immediately and then you can drill down, get as a specificer as vast as you want. And if we don't have a built-in panel, though we do have hundreds of them for all the most common use cases, we do a lot of really advanced queries so you can design your own panels, HTML support, you can go to outside data sources, really, really get in there and build exactly what you want based on all the data we're saving for you. That's it, exactly what I want. That's it. So depending on what environment I'm in, I'm going to care about certain things and not care about other things. And there's certain metrics I might be really important to me and other things in other environments that maybe I don't care about. But that you just said the magic words for me. I can build what I want. Exactly. And if you don't have the pre-built panel, I can feed that data in. Are you going to be able to get that and graph it for me? Super customized, Lewis. So can you walk us through a couple of customer use cases? You know, we've got a US retailer

34:41who has a very large environment, 10,000 stores across the country. So a lot of challenges. They wanted to map their entire environment. So you can kind of plot flat longs against devices in Statsika. And we were able to kind of work with them and create a proper 50,000 foot map view of their entire environment. It was fully dynamic. They could go right to the store level if they wanted. And they plotted every critical device, including their primary and secondary router at that location. And they could see for their management users, for their network team, things that were happening in real time, especially things like weather events, power outages, as well as planned firmware, kind of rollouts and changes that were impacting stores. And in addition to seeing it kind of happen in real time and live, they could then use that data historically and look at all which stores were impacted for the longest when did services come back and restore? And they could then kind of use that information

35:43for future kind of planned changes and another seasonal reporting. I think I'm Andrew here. So you just said 10,000 stores. This is a huge retail environment, right? Yeah, correct. Yeah, this is okay. We've got some of these customers, yeah. And this is not just red light green light. I heard a lot more kind of information that I can pull in here. So I shouldn't be thinking of this as like, oh, you tell me when things are down. There's a lot more going on here. Yeah, correct. Because we kind of have a lot of flexibility in how you choose, you as a customer choose what metrics you want to kind of focus on and be a kind of a priority. Yeah, stepping away from just the traditional up, down, you can actually be specific about maybe looking around trip time or IPSLA or health metrics or even I've worked with some customers on an aggregate of those that kind of creates a bit of a health score. So there's kind of flexibility to do that type of thing right within the platform. And you even mentioned outside events that aren't necessarily rated to network performance like weather.

36:44Like I'm from the US East Coast and we just got hit with a couple of big storms over the past month that did impact like shipping and delivery and the ability to get stuff to people and all that. And I can imagine that could be useful information particularly in a retail environment. I've had customers that have their network environment on a map kind of side by side with their weather events as well as the information that comes through HTML panel from their different power suppliers that actually gives them real-time information around outages that are occurring that impact parts of a city. So you can correlate a whole bunch of really nice kind of pieces of data alongside that helps to kind of make a quicker decision for your network team and it also helps with things like management to just kind of calm a management user to say this is fine, this isn't a network issue. This is kind of something that is unrelated to us. Don't stress. Yeah, that mean time being essence by just putting like, there's six feet of snow. So, and the power's at.

37:45Oh. Yeah. So there's lots of network monitoring and network performance monitoring products out there. How would you differentiate that, Taker? So there's a lot of tools out there. There's open source solutions. You know, every vendor now has a platform. So there is a lot of options out there. I guess going back to that point around what our DNA is. It's really around data quality. The efficiency in how we collect kind of and store data and then the focus of what Statsik is all about. So accurate data for us is everything. You know, we value data kind of quite a lot and we value our customers data a lot. So that 60 second poll cycle that is kept unavvaged, that's that's huge for our customers. And that's why some of them have had Statsik for 15, 20 years because they go out to the market. They look for a solution that can replicate what Statsik is doing for that kind of source of truth. And it's not there.

38:46You know, they can't replicate that, especially at scale. You know, with this big data collector, highly efficient platform that doesn't hurt users by scaling up. You know, so we don't have to have lots of heavy tuning and complicated licensing to kind of make Statsik work. So by that, in its very nature, you know, that's kind of one of the differentiators that kind of makes it an administration, simpler kind of solution, and also a lower total cost of ownership solution. And one of the important pieces that there's a lot of solutions out there and Statsik is not your, not replacing your seam. It's not trying to be your ticketing tool. It's not trying to do everything. It's very focused for network teams and networks on creating and providing clean, accurate metrics that can actually power proper decision-making kind of processes. And that's why even this morning, I'm on a call with another customer

39:47and they're just saying, Statsik is our source of truth. We can refer to it, we can rely on it. It's trusted within a team and they can then use that information to kind of drive what they're doing during a day, a week month, a whole year, for example. Now, you mentioned along the way licensing could talk about that. So, okay, how is the product priced? Great question, Ethan. But everybody wants... I mean, obviously. That's why we'll put it at the end. Look, if your listeners take nothing away from this, Statsik is deliberately uncomplicated. It's a very kind of focused solution on doing what it does in the network, monitoring space really well. And we extend that simplicity to our pricing. It's a per-instant subscription with a device count. So, how many devices you want to monitor across your network? That's what makes up your kind of licensing model.

40:49So, that starts from $6,500 for 250 devices and then scales up there into what I spoke about before, tens of thousands of devices from a single Statsik, or instance. And what about... Are there add-ons, or especially, specialized modules? Because you mentioned integrations with other platforms and products. Yeah, then the kind of modules that we spoke about, things like Cisco and Maraki, Cisco ACI, and then also our observability appliance for that distributed remote polling. Those are kind of add-ons if those relate to your technology and your environment or the architecture of what you need. So, we can kind of bolt those on for additional cost in and around that licensing model. Okay, but $6,500 for 250 devices out of the gate and then you scale from there. Correct, yeah, correct. So, it's not there to blow up your IT budget. It's there to really... In a lot of ways, complement, but be this powerful solution that can really scale properly with you.

41:51I think I know what you mean by devices, but just to qualify, because I know a lot of network-moderating platforms are prized by device. You mean a device that's got a switch with 48 ports. You're not building me per port. Yeah, correct, yeah. Like, there'll be a device and it'll come with a certain number of interfaces. But what is kind of important, I guess, is that device count. So, device count. Okay, yeah. Not the element count. That's the way it's built on some other platforms. The element and an interface is an element. That's not what you're doing. So, we mean natural devices. Yes, okay. Yeah, nice. So, no, there's no sensors. There's no kind of... There's no flows. You know, we're trying to keep it kind of simple. And it is. So, the other big question then besides pricing is, can listeners play with it? Look at it. Try it for themselves. 100%. That is what I always recommend people do. We offer a 30-day free trial. It's fully featured. So, we're giving you everything in the core product

42:54when you do that trial. And it's just like everything else we talked about today. It's super simple to set up. There's no credit card. You don't have to get on a sales call. And the trial is set up to show you value very quickly. You can deploy and see live data fast. Then you can jump into those pre-made dashboards I was talking about to find patterns, gaps in your capacity and just start producing actionable reporting right away. Usually most users can be up and running within a couple of hours. Install it. You run a discovery to find your network devices, putting your credentials. And it's pretty much up and running from there. And then you can really drill in. So, yeah, that's a good way for listeners to validate in their own environment. Whether it's a good fit for them and how quickly they can detect and explain issues going forward once they have stats. Seeker is part of their daily process. You guys must like engineers. No credit card in those sales call.

43:55It's like it's almost like you understand how engineers think. I don't want to talk to anybody. Let me try the thing. Please just lean me alone until I'm ready. 100%. I definitely come from the engineering side before. I spent a long time building networks and I was a stats-eeker customer for a long time. So I know I know that side of it very well. All right, well, we're at time, but hopefully we've wet people's appetites to go and explore this for themselves. If they want to go find out more or get the download, where should they go? statseeker.com slash net ops has everything you need on there. Just check out that website. You can sign up for the trial. Get some more information on our features, system requirements, all that. It's all documented statseeker.com slash net ops. All right, that's statseeker.com slash net ops. We'll also have that link in the show notes that accompany this podcast. And if you do reach out to statseeker, we'd love it if you let them know that the pack of pushers sent them your way, because that helps us. Because sponsors do make what the pack of pushers do possible, so we can offer high quality,

44:55deeply technical content for your professional development for free. That includes more than a dozen technical podcasts on networking security, IPv6 DevOps and more. We've also got an industry blog two weekly newsletters, our community Slack group, a YouTube channel and IRC group. You can find it all at packopushets.net. All free, no login required. You can hear us on Spotify, find us on LinkedIn and read us on our podcast. And last but not least, remember that. Too much networking would never be enough.