Skip to content
PodSubs
ShowsPeopleGuestsJobsNewsEvents
More
BlogPodCred
Menu
ShowsPeopleGuestsJobsNewsEventsBlogPodCredNetworksSign in
Join nowSign in
  • Shows
  • People
  • Jobs
  • News
  • More
    • Guests
    • Events
    • Blog
    • PodCred
    • Networks
    Join nowSign in

Threat Talks - Your Gateway to Cybersecurity Insights

Threat Talks - Your Gateway to Cybersecurity Insights artwork

MDR Has To Stop Mopping The Floor

Sep 29, 202617 minAudio
ShareXLinkedInFacebookWhatsAppEmail

A CISO got breached and could not say how. No entry point, no list of what walked out, nothing to take to the board. He paid the ransom blind. Then he bought MDR. The next time, he had all of it. Who clicked, which data was touched, when it left. He still got breached. So what did the M in MDR actually buy him? Lieuwe Jan Koning, Co-founder & CTO at ON2IT, sits down at RSA Conference with Nicholai Piagentini, Technical Enablement Engineer at ON2IT and twenty years deep in the Bay Area security scene, to separate what MDR vendors pitch from what MDR delivers. Buyers think MDR is one product. It is five jobs: collect the logs, enrich them, put an experienced human in the loop, remediate, and keep it all retrievable. Most vendors do two of them well, and only inside their own stack. Three questions tell you which two. All three are in the episode, and in the description below. Timestamps (00:00) - The MDR reality check: expectation versus practice (01:33) - What buyers think they are buying (03:35) - Single pane of glass, and why nobody has one (06:40) - Why no vendor is paid to support a competitor's logs (07:42) - Log retention: when everything becomes hot storage (11:06) - Three questions to ask every MDR provider (13:20) - Data sovereignty: who operationally controls your logs (15:02) - The future of MDR, and why detection alone is cleanup (15:36) - The CISO who saw everything and still got breached The three questions In practice, what does your platform do with a log it did not create? Ask for the demo on a competitor's telemetry, not your own. Is there tiered storage, or does every log stay hot for the life of the contract? When I escalate, who picks up: someone who knows my environment, or someone reading an LLM's answer back to me? Key topics covered What "managed" means to different MDR vendors, and why a human in the loop is rarely part of the deal Why EDR, NDR, XDR and DDR all get reskinned as MDR, and what that does to scope The economics of vendor data lakes: every log priced as hot storage, with no tiering to warm or cold Commercial incentives that keep cross-vendor log support shallow, and how to test for it before you buy Data sovereignty and jurisdiction as a procurement requirement, not a European footnote Related ON2IT Content & Referenced Resources: Threat Talks: https://threat-talks.com/ ON2IT Zero Trust as a Service: https://on2it.net/ AMS-IX: https://www.ams-ix.net/ams Subscribe block Subscribe to Threat Talks and turn on notifications for deep dives into the world's most active cyber threats and hands-on exploitation techniques. πŸ”” Follow and Support our channel! πŸ”” β–Ί YOUTUBE: / @threattalks β–Ί SPOTIFY: https://open.spotify.com/show/1SXUyUE... β–Ί APPLE: https://podcasts.apple.com/us/podcast... πŸ‘• Receive your Threat Talks T-shirt https://threat-talks.com/ πŸ—ΊοΈ Explore the Hack's Route in Detail πŸ—ΊοΈ https://threat-talks.com πŸ•΅οΈ Threat Talks is a collaboration between @ON2IT and @AMS-IX

Audio and video play straight from the show's own hosting. PodSubs stores and serves no media.

Transcript

From the show. Tap a time to play from there.

Today we're doing an MDR reality check. If you procure MDR services, what do you expect, what you want and what did you actually get in practice? Welcome to Threat Talks. My name is Lieuwe Jan Koning. And here from the head office at Venable in San Francisco at the RSA conference. Let's shed some light in the mystifying world of MDR vendors. Let's get on to it. Welcome to Threat Talks. Let's delve deep into the dynamic world of cybersecurity. I'm joined here today by my colleague Nicholai Piagentini.

He is a technical enablement engineer here based out of San Francisco. Nick, welcome to the show. Thank you for having me. Great to have you back actually, yeah. So you've been in the space for in, well, around the Bay area for decades, actually. Yeah, over 20 years. Over 20 years. Yeah. You've seen many, many vendors and, well, your kids go to the same school as all those other people that work everywhere. So, yeah, you have a very good view on what's out here in terms of vendors, in terms of profiles of MDR, XDR, MDR, whatever it is.

MDDR, MDR yeah, what they do, what they don't do especially. So yeah, you're my first choice, actually, to talk to you about what customer expectations are when procuring an MDR service and and what the reality actually is, and if there's a gap. So and we'll leave the audience with a few pointers what to what to look for and what to be careful about when selecting one. So yeah, what would you say is something that organizations really expect? Yeah. So I think it's really interesting in that MDR

can mean a lot of things to different vendors, right. It's not how would you define them. The standardized her managed detection and response. Yet we often will see tools that are not managed really in any way and don't actually cover any response. Right. Maybe it's just detection or that. And so as a result, it's not really a clear tool. I think when customers go through the booth, talk to somebody selling some kind of MDR XDR solution, they think what they're getting is a single pane of glass, one place

to consume all their logs that could be related to security in any way that can apply logic and analysis to them, enrich those logs, preprocess them, act as an extension or force multiplier to their SOC, right, and then provide automated or extremely clear remediation. Help them with any kind of maybe auditing right or compliance, and ultimately provide them with a data lake of their logs for look back right.

Or for threat hunting historically. And so this is "M" for many is it doesn't mean managing the outcomes of the events or like managing the threat that is that is there and remediating it. Or does it mean managing the fact that logs are being collected? The answer is yes because different vendors it means and or question. Yeah. I think what we find is that many of them do not see the management as a human in the loop, which really, you mean is supposed to mean expectations.

You mean or what? What's being offered in the expectations, right? They see it as just a tool that pulls these logs together and provides this enhancement. They do not envision specifically maybe another SOC engineer as part of it, which is unfortunate because that's really the best way to deploy it, right? You need a human in the loop for that, at least for now. Now, in current state of AI, certainly, absolutely. Now with the current state of AI, no. Okay. And what about the scope of things? I mean, because MDR typically what we see in Vietnam is that is an endpoint solution that now has an MDR type surface on top of it.

So it's EDR endpoint detection and response becoming MDR or network detection and response becoming MDR or XDR becoming MDR. Yeah. Or data detection in response. Yeah, yeah. Every vendor skins it for what they do, right? If every if you only have a hammer, everything's a nail, right. So whatever the vendor does that becomes kind of that first letter there. But it is always pitched as a single pane of glass. Right. That term gets so overused. But the reality is so the expectation is that the scope is everywhere. So that it system that's the endpoint, that is the network that is the OT as well

probably. Well, I mean, very few vendors would actually say that because they don't cover it. But yeah, the implication is that this should be the tool you use to consume your logs, apply logic to it, maybe AI, right, and have a clear outcomes, right? Adoption in noise, all of those features. But the reality is it is very hard to find a tool that does all that you need in one. You'll end up with multiple MDR, right? Whatever. Whatever you want to put in front of it was really removes the advantage of it.

Yeah, because then you end up with multiple. But there's a thing called platform mission for that. Right? It's a made up word. But yes. Yeah, yeah it's the idea of but well that's a term that can be thrown out that implies things all working together cleanly, natively. No one has accomplished that even within large security companies, right? Companies that cover many different aspects of cybersecurity, they are really unable to pull all of their own logs together to one place. Why is that? You mean you think?

I think it's because many of these logs have very different context. Many times they've come, so it's always been hard product or exposition. That's a different thing. It's an integration problem then. Yeah. You know there's there's the idea that I can consume all the logs. I could theoretically create a tool that takes any log I send it. Yeah. It's called a sim, right, right. Just throw everything at it. Yeah. Can I provide that same level of enhancement and intelligence to every kind of log that comes in? And if we look at the industry, the answer is no. They have very powerful tools around their own vendor ecosystem of logs.

But when you move beyond that, you get very low fidelity information coming it right. They're unable to contextualize all the fields the same way they could from their own logs, and they are unable to apply the same kind of rigorous analysis to it. Do you think it's because it's really hard to do the analysis? So say, say you have a firewall for brand A and for brand B and then the, the MDR servers or the data lake is from one of those two vendors, and they support the other vendor less indeed. But they could probably. Or is it is it that hard to think or is it is it a matter of trying

to convince your customer to use all the products in your portfolio? A platform could also be making one vendor and everything, so say for that as well. I mean, it's easier to deal with one vendor. Yeah, the one throat to choke. I've heard that phrase. It's definitely true. The real problem comes right? That when we have this idea of how can we work with all these logs, a company is not heavily incentivized financially to provide detailed support for their competitors tools. Right. There's always that need that desire for your field to upsell, right. Land and expand.

Okay. You're using our firewall, but somebody else's endpoint. Sure, we'll support that. But look what you'll get if you remove that other vendors endpoint and come to our endpoint. Look how cleanly we stitch these logs together. Right. If they could provide that clear visibility regardless of the vendor and have nowhere to expand. Right. And as from a business perspective, from a security outcomes perspective, that's great. Everybody should be doing that. That is the way we should do things. From a business unit perspective. No one's going to do that.

The reality isn't that everything I mean, you have acquisitions, there's point processes. You sometimes need certain features or lagging behind in a certain feature. So you'll reality is that you do end up with different families. You have to no one has a full stack across everything. No one, even a really big audience. Hey, what about lock retention? I mean, one of the things that I hear a lot, at least, is that you want for forensic, for example, you want long term storage. Yeah, yeah, yeah, we need three months. We need seven months. We need a year. Right? We need seven years. Whatever it might be. I think what amuses me the most about this right

is within our industry, in the concepts of storage. We have really great, mature ways of tracking hot storage, warm storage, cold storage that can become very cost effective, right? Cold storage S3 is trivial, right? The cost to store stuff that's not quick to read or write to. Then we get to your MDR or whatever it is. Now all these logs are going to a vendor data lake. They're all effectively hot storage because they want to be able

to search them, perform tasks on them. So you're looking at the highest price for terabyte of storage. There's an extra add on because you're not just buying it straight from whatever storage vendor, right. You're buying it through your MDR product. Right? So now it is doubly expensive for terabyte, and you have no way of of pushing it off to warm cold storage, let alone data. Salvatore. Right. So we get to that afterwards is so expensive to consume data into

because it's always more logs that you have than that you anticipate that your that your ROI tool recalculated. And how often do you have to do a look back for seven years or seven months or even three months, like, do you need to spend this if you just you if you if you discover that there's some kind of Chinese guy already in your network for three years, you might want to. Yeah, yeah, I want to see what I did exactly. But where do you want to spend that money? Do you, do you store everything with that in mind? And you end up spending so much money on storage, right. It's really one of the areas that balloons the cost of these solutions, right?

When you start looking at the storage, despite having very inexpensive, high performance storage options available. Right. But they are outside of that vendor ecosystem. So you mentioned a couple of things in terminal expectations. It's a single pane of glass is important. One cross vendor that is an expectation or a requirement, a scope that's broad for not just networking or just endpoint, but actually everything. Because that's if you buy something you get you get a single pane of glass dashboard.

You expect it to have truly organization. You don't have to go to another dashboard and have another subsection of it. Anything else there? I mean, yeah, there's the idea that it will provide some turnkey sort of compliance and remediation. Right. So I think a lot of times the idea of playbooks gets tied in to this. We detect it. And now we can trigger these playbooks. And certainly walking through the floor of the expo these last days, the stories that are told is always about, oh we do this out of the box, very turnkey.

Everything works together. As you dig into that, you find that there are very few turnkey solutions out of it. And you ultimately it's important to be turnkey. I mean, if I don't know, station needs to be. Obviously it was very small organizations. Maybe they don't have the expertise to build out the playbooks. We should definitely see generative AI or other tools, right. Making things easier. But it's certainly what sold right. And if someone's buying their MDR, basically the thought that they will have automated tools to handle all of their situations and compliance needs,

they're going to be sadly, sadly disappointed. Okay. Well, let's let's let's talk about a couple of requirements then that you should have some things that you questions that you need to ask when talking to an MD provider. I mean, because you say it may be costly to have long term storage, it may be narrowly scoped, etc.. What do you think? What would be your top three of things that you need to be careful, given the space of the I mean the obvious ones like like you need to respond to a known

a block list or high that everybody has that that is not a not something that narrows the field. A little bit. But yeah. Yeah. So I'd say first of all understand the difference in performance and features. When you look at a vendor ecosystem log or the external log to that. Right, you were going to have different performance on those two. They're going to have enrichments, different detail. How much different what is it. You should actually see it. Yeah. You just test it out and then and then on purpose

and make put a product in a log, sort in and see what happens. Is it able to link it to other things. And it stitched that together with other logs that are meaningful. And then I think the second thing is do you have tiered storage. Do we have this hot warm cold? Unless you are fine with one month or yeah, you're fine for a very short window, right? Yeah. That would be okay. And then I think maybe the one thing we haven't really mentioned back to that M is there a human in the loop option?

Yeah. To escalate things to who do I ask you to. Yeah, it's my responsibility. My SOCk. Do you have a SOCk that can help me? Is there somebody in the loop? And is that person experienced? Right? What kind of person am I reaching out to? Is there someone who's just going to put my question to an LLM and then read those answers back to me, because then I could do that, right? Is it someone who knows my environment? And these are things that are outside of the technical side of it. Right. And many vendors really don't go far beyond the technical right. They don't have any concept. It's more of an alert copy paste machine, more or less. Yeah.

They're like, we don't know what you're doing with your SOCk. You have your own processed air. Go ahead and use it. But then that makes this not your single pane of glass, not your one place to go. Right. But just another dashboarding tool that you have to use. Okay. You mentioned a little bit data sovereignty. Yes. And I think this becomes more and more important. But that's a very European or Canadian thing I'd say it certainly not a U.S. thing. Right. Though initially it always has been right, which is why it's not as common a term here. So we were beginning to see states implementing requirements.

Right. Federal government has certain requirements. Data sovereignty doesn't necessarily mean the US government can't see it. It means that if the US government is a customer, they control who sees it. Right. And this is going to become so important because so many of these tools now consume these logs into the vendors cloud, where they can apply all you don't know where the cloud is and they can, they can and country but which yeah, don't know where it's going to be. It's going to have your fiction is actually important. Jurisdiction is important.

Who operationally controls it. Is that your vendor or your vendor. Vendor. Exactly, exactly. And we can see it. And that that can no longer be a question mark. Right? I think for a long time you could get by without that in the US market. Right. Because it wasn't a question people were taught to ask. But I've been seeing more and more of that question coming up in the US market. Right. People are recognizing that they need to own their data, right? They need sovereignty over their logs, which represent very proprietary data about the organization. Yeah. Especially if you indeed log everything.

Yeah. Which we want to do. Right. Like the the value of it is in how much it consumes. Right. Yeah. I don't need it to be really good with a small subset of logs. That's uninteresting to me as a security practitioner. Right. If I'm going to have a place that's looking at my logs, it must look at all of them. So we should also talk a little bit about the future of MDR, how it is changing, because so far what you've been saying is a lot about collecting logs, responding to things that are broken. And although the M is

assumes a little bit like we're going to do something about it, not really. What we see in for a pure MDR product is more is really like a like a thing. Yes, indeed, it will tell you how things are bad. And I was joking. It gives you the perfect report like, hey, this is this is why you should be fired. Because. Yeah. Oh, I thought I saw the whole thing. It was terrible. They got everything. I had a conversation. I may have shared this story before, but I remember talking to a CISO that said, hey, we got breached and all my.

And we didn't know what hit us. I mean, it was out there, we had to pay ransom and we didn't know how they got in, what data they had. We knew nothing. So then we invested in an MDR product and now we actually saw everything. The next time it happened, we know who clicked on the link. We we know which data was accessed when it was happened, how much it was everything. And I asked him, but you still got breached, right? So but let's, let's, let's talk about that next time in a follow up episode. What we believe that the future of MDR, because it's a bright future

might be it might be dark now, but the future, the future is bright. I'll hold you to that for next time. Thank you very much to our audience. Thank you very much. If you don't want to miss that next time episode about what the future of MDR. Press the subscribe button. And there's also a like button there. We would appreciate you press the button as well because it helps us spread the word. But for now, thank you very much and hope to see you next week. All right. Thank you for listening to Threat Talks, a podcast by ON2IT cybersecurity and AMS-IX. Did you like what you heard? Do you want to learn more? Follow Threat Talks to stay up to date on the topic of cybersecurity.

Open the transcript file

Hosts and Guests

Nobody is credited on this episode yet.

Credit the people on this episode

Be the first to rate this episode

More episodes

All episodes
  1. No CVE, No Patch: The Phone Call Your Phishing Training Missed

    Nov 17, 202616 minAudio

    Your card never leaves your hand. It still pays out at an ATM on another continent. All it takes is one convincing call from "your bank." The victim installs an app, taps their card on the back of their phone and types...

    Credit the people on this episode

  2. Why Can Your Printer Talk to Your Domain Controller?

    Nov 10, 202615 minAudio

    No patch, no defense. That's how most teams think about zero-days. So what decided how far an AI agent got inside the Dutch Institute for Vulnerability Disclosure (DIVD)? It came in through two unknown flaws in the...

    Credit the people on this episode

  3. How to Prove to Your Board That Zero Trust Is Working

    Nov 3, 202631 minAudio

    Your board wants to know one thing: are we safer than last year? Alert counts can't answer that. Twice as many events can mean twice as many attacks, or a sensor that finally works. Lieuwe Jan Koning, Co-founder & CTO...

    Credit the people on this episode

  4. Citrix NetScaler Zero-Day: Three Weeks Without a Patch

    Oct 27, 202617 minAudio

    Citrix left customers exposed to two live zero-days for three and a half weeks. Rob Maas sits down with ON2IT CTI specialist Lars Nobelen to break down how attackers exploited CVE-2026-88771 and CVE-2026-88772 on Citrix...

    Credit the people on this episode

  5. Stolen iPhone, Tracked Live, Thief Caught

    Oct 20, 202631 minAudio

    Fifteen minutes after Lars Nobelen's iPhone was stolen at a scrapyard, it was already useless to the men holding it. An hour after that, police had all three of them in custody. Lieuwe Jan Koning talks with Lars...

    Credit the people on this episode