Why Data Recovery Isn’t Enough for Cloud-Native Apps
A third act in data protection
Kedem has spent 25 years in backup and recovery. He co-founded Kashya, which EMC bought in 2006, and then Zerto, which HPE later acquired. After two and a half years off, he kept hearing from former customers that they had never run an end-to-end recovery test of their cloud-native apps. Meanwhile, those apps are where most new development happens. That gap led him to start Dreeze, his third company in the space.Why cloud-native recovery is so hard
A modern application is more than its database. It depends on hundreds of cloud resource types, networks, secrets, permissions, Helm charts, Lambda functions and third-party services. Yet most teams back up only the data. So when something breaks, they face what Kedem calls the recovery scramble: a war room, a hunt for keys and infrastructure rebuilt by hand. As a result, real testing rarely happens.How Dreeze handles cloud-native recovery
Dreeze first discovers every resource, groups them into applications by criticality and maps their dependencies. Next, it protects both the data and the full configuration. Finally, it orchestrates recovery end to end, which makes regular testing practical instead of a once-a-year table exercise.Why AI makes cloud-native recovery urgent
Kedem notes that AI makes attackers more capable and human mistakes far bigger, since an agent can act on an entire system at once. In other words, failures are a matter of when, not if. Being able to recover any application quickly lets companies take those risks safely. For Kedem, reliable recovery is the safety net that makes faster innovation possible. Dreeze launched just a week before the interview. To learn more, visit dreeze.com.Transcript
Hi, everyone. Welcome back here to Techstrong TV. My next guest is Ziv Kadem.
Ziv is co-founder and CEO of Dreis, and we're going to hear all about that and data recovery and cyber resilience and a lot more. But first, let's say hello to Ziv and meet him a little bit. Ziv, welcome to Techstrong TV.
How are you? I'm good. Thanks for having me here.
My pleasure to have you on. So Ziv, as I mentioned, you're the co-founder and CEO of Dreis, but let's hear a little bit of what brought you here. How did you get to this place in your journey?
Look, some of my life led me to the last 25 years of doing data backup and recovery. My first company was a company called Casia. I started coming out of the Israeli army.
We did that for physical servers for open systems, right? That company got sold in 2006 to EMC. It's EMC RecoverPoint product today.
And then virtualization really started taking over IT, and everything got virtualized, and we needed new solutions. So together with my brother, Oded Kadem, who was also with me at Casia, we started Zerto. And Zerto grew.
We grew to triple digits of revenue, 10,000 customers worldwide, some of the biggest names. And then that got acquired by HPE- Yes ... and that's HPE's Zerto product.
And so leaving Zerto, I thought I took some time off. I thought that I will never do backup. I thought two companies are enough for someone.
And then speaking to some of my previous customers, we found that there was a huge gap in the next generation, which is the cloud-native applications. Yeah. And cloud by itself didn't change, it didn't break backup.
But the modernization, cloud-native, it created a whole new thing. So literally, I spent two and a half years on vacation and relaxing after 15 very intensive years of Zerto and just seeing that big pain, the big gap of the difference between what you could do in the virtual machines and hearing several former customers telling me they never had an end-to-end recovery test of their cloud-native apps, which is really the future. That's where 80% of their development is.
Absolutely. It's funny, Techstrong's part of Futurum Group now, and a sister company in Futurum Media. Futurum Media is the group of media companies in Futurum.
So it's Techstrong and a company called Tech Field Day and Visible Impact. I was at a Tech Field Day security event three weeks ago. HPE was presenting, and they actually spent most of their time, not most, but a good chunk of their time talking about Zerto and their plans on where Zerto is and what they're doing and everything.
So it is alive and well there, and it really is the backbone, right? It is their data recovery solution. It was very, very interesting.
But of course, you're right. Cloud native is the stack, the compute stack that most people are building on today. And I'll add, as I mentioned to you, I was out at KubeCon last week in Salt Lake City here, and I did a presentation.
I did a panel on Monday at an event we did, and it was that the cloud-native stack is the AI stack. So not only is the cloud-native stack the default stack for, so let's call traditional computing, distributed though, on this Kubernetes and everything that come containerized and everything, but all of our AI applications, and most AI has Kube at the center of it, right? It's Kube that's orchestrating, and it's that stack.
It's all serverless platform as a service. Yeah. That's a modernization stack, right?
That's modern. All the AI is going on the modern stuff, all the agent, everything. Absolutely.
I remember, Ziv, three, four years ago, people saying, "Well, what's going to come after Kubernetes? " Then AI came on the scene. " Nope.
Kubernetes kind of absorbed it, and now it's AI. But it is a very different, especially from a data recovery point of view, it is a very different motion. It is a very different kind of how do you got to do that?
So I would imagine this is a whole... I'm frankly surprised no one saw it before. It's a relatively new problem, right?
" That was all just lifting and shifting. Then people are now starting to take the next phase, and now we're starting to see that modernized application, that serverless platform as a service, Kubernetes. Sure.
And that's really where you get the real value from the cloud. But again, today, so their need is bigger. In most of the customers, they're 20% virtual modernized.
And that's new, right? A year ago, they were 10%, right? Yeah.
And it's growing. So it's starting to hit now, but they're starting to look forward-looking and saying in three years it's going to be 60%, and then it's going to be a real problem for us. Sure.
It's funny, we started a cycle of digital CXL four years ago. Three, four years ago, because we saw this transformation happening. It wasn't lifting and shifting.
Cloud washing your old application was no longer enough. You really had to re-architect it, right, for a cloud native kind of stack with microservices and these kinds of things. Yeah.
And it really started gaining momentum. Like you said, 10% to 20%. The numbers I saw are 15% to 20, 25%, but it's moving.
But you know how life is, right? And how technology. This is why I love doing technology.
As that's happening now with AI, and because of digital sovereignty and sovereign AI and all these other things, people are now saying, "Wait a second. I may want to move some stuff back from the cloud to my data center, even my server closet. I want to run it locally.
I got to own my IP. " But I still want that cloud-native stack, right? So now I want to run Kube.
A lot of people were moving either Kube on top of a VM or Kube on serverless, as you say, and all of these things. " But I want it all, right? Because this is just the insanity that we deal with.
Just as one trend really picks up speed, the next trend- Mm-hmm ... starts getting in there. So talk to us about Dreis a little bit.
How does it help with all this? So really when we spoke to customers, enterprises, we realized they have four major gaps in their ability to really-- They would like to be able to revise their applications when something goes bad. But really, the first thing is that they don't know how to recover.
Applications have tons of resources, more than 700 types of resources in AWS. Then you add Azure, Google, OCI, and then these resources have dependencies between them. They can be complex, they can be circular.
So you don't really know, okay, now how do I bring an application back? Next, you go to the protect. They only protect the data.
They back up the data. They don't back up all the configuration. The application has not just the data, it has the infrastructure, the VPCs, the networks, the secrets, the security, the application there itself, the Helms for the containers, the Lambda, that's where you deploy the agents, and then you've got third-party connectivity.
And these modern apps, it's all built about not rewriting, but using third parties. And you need to get all of that right for the application to run, right? So when they only protect the data.
Then the next point is how do they recover? We call it the recovery scramble. You set up a war room, find a process, spin up the infra, find your best DevOps people, hopefully trying to look for the keys for this, for the API keys, for permissions, and you get one thing wrong, you go back to the start, right?
And because it's so complicated, right now when you have an incident, it's going to be all hands on deck. But for testing, then testing becomes very limited. You don't test enough.
You test limited tests. Database recover the records, but can you actually run an application? So they really see that they run the risk of getting the data back with the application still dead.
Now, Driz is building that modern native cloud backup and recovery solution. Takes care of that entire thing. So we first discover everything you have.
We classify it into applications by criticality, map the dependencies, create that automatic full recovery plan. Then we protect everything, not just the data. Of course, we also protect the data.
That's the key. Without data, you have nothing. But we also protect all the infrastructure configuration, all the application they're running, and really create that full flow.
Then we orchestrate it end to end, so you don't depend on manual processes, on manual documentation, on specific people. And because it's so easy, now you can test it as often. You can validate through testing.
Absolutely. And right there, as long as I've been in IT, and I've been in tech 35 years, maybe more even, here's my experience. Someone got the data recovery solution.
I'm going back to the days when we put it on tape and everything. Someone bought a data recovery solution, and they said, "Okay, check. " That person left.
Right? And then a new person came in, said, "Ah, I don't got to worry about it. " And no one ever tested the damn thing.
Or if they did a test, it was cursory. Oh yeah, it's still up and running. And then the stuff would hit the fan.
You know what I mean? Then something would happen. We'd have an incident, and then we find out just what you're talking about.
Yeah, we got the data. We don't have the configuration. We can't re-set up the database.
We can't restore this. And even if we can, it'll probably take us two to three weeks minimum. Right?
And in today's world, that's just not acceptable. So- It's exactly that. It's hours turning to days and weeks of something that could be automated.
And the worst thing is, right, we know today people are speaking it's not if, but when. And you spoke about AI, right? When you think about the if, it used to be hackers and human errors, right?
Now, the hackers are 10 times more powerful. The human errors are 100 times more powerful. More powerful.
Right? If I would give you a task to close 100 ports, then you would close one, you would close two, and then someone would yell, and you would stop. Now, if you tell Claude, "Close these 100 ports for me," if you're lucky, it'll only close these 100.
Right. So you've got blaster it, and everyone is deploying agents, and these agents have access to a lot of critical things, and they can do damages at agent class radius. Right?
So if you're looking at failure, I'm only saying cyber made failure right not if, but when. AI is going to make it how often? So to me, it's AI scale.
It's the scale of this. Yeah. We can't-- And I think there's a third dimension that bears putting in, too, and that is when we're talking about cloud-native infrastructure, a lot of it's ephemeral, right?
The average container is a couple seconds. Right? Depending on the app.
And so when we talk about being able to restore it, I don't want to restore it from last month. I want to restore it from last minute or last hour. Right?
So the idea of keeping that current is another scale issue, right? That's a huge scale problem. Yeah.
It's changing all the time, right? Yeah. That's really the problem, right?
It's all a problem of scale, right? When you looked, when it was virtual machines, a big application would have 10, 15 VMs, virtual machines. All of the complexity, all the networking, the security, the identities, everything was packed into these virtual machines.
Now, that's the reason people are moving away, because you would change one small thing, everything breaks. Yeah. But now, the database only holds the data, right?
Everything else that's to do anything meaningful with that data is stored elsewhere now. And can it be recovered? Yes.
Can it be recovered before it causes a lot of damage? Right? Is the failure, okay, just a bad Tuesday, or are we going to have to go and do damage control?
That's a different thing. And how confident are you that you'll be able to do it, right? Is it going to take another-- Right today people feel very exposed, because when they say, "Okay, we've never really practiced it.
We've done table talks assuming it will do. " It took me a week to get the us ready. Exactly.
" It's a big problem, but it is a solvable problem. I hope it's solvable. " We certainly live in interesting times, right?
So just in this past week or, yeah, within the last week, we had the refrigeration systems in France were subject to a cyber attack. We had this ground stop in the US, all airplanes coming to a halt. I sat on a runway for four and a half hours.
We had Microsoft Outlook- Microsoft. yeah, went down for days. Yesterday was the banks here in the US.
I don't know if you saw Chase and Wells Fargo, a bunch of them- No ... had issues. And I'm thinking, I'll be honest, I was thinking to myself, coincidence?
Is this some sort of low-level terrorist or state-sponsored kind of attack, or is it that our technology is becoming so complex that the AIs and everything doing all of these things are just destroying our resilience? Right? Our ability to recover quickly and fix these things is just we haven't-- We're losing.
Look, I think it's a point-- I don't think we're losing, right? It seems like the world is moving forward, but it's definitely that there are more forces going into creating those failures, right? And it's mostly bad forces that got much better.
Right? We know that AI is much better at finding a security breach than at closing- Yes, interesting ... all the stuff, right?
So it's really, it gives an advantage to the bad actors. " Right? I think it's inevitable, especially, like we said, a lot of it is self-inflicted.
It's going to be from agents or from your employees. When I run a Tell Cloud to do something, there's an agent running with all of my permissions. And we're seeing that people are letting a lot of these controls more loose, which is great for productivity.
It's not great for risk management. And I also think that that's going to create a competitive advantage for the companies that-- I think there's no question today that if you go and release these agents, the most free will get a lot of efficiency- Yeah ... if they don't kill you in the meantime.
So I think that, like I said, so if you're not as worried about that, that you can recover from anything from a single app to everything, then you've got advantage. You can take these risks. Because companies are taking risks.
Like, with AI, everyone is taking risks that they never thought about taking. Some of them are probably backfiring. We once said that we launched a week ago, and everyone goes, has outages that week to support our launch.
So- It was nice of them, huh? Yeah. They told me to send the thing to Microsoft, and then everyone else started breaking.
But I think we're going to see that, right? Failures are going to happen, people, because otherwise, you're either going to close yourself with all these controls and stay behind, and so people will have to take them, but they'll have to make a manageable risk, not just letting them wild, right? Because- Sure.
So Zeev- Like doing stuff with a second net. We're the second net. It's a good way of putting it.
Zeev, we're over time already, but for people who want to get more information on Driz, what's the best place to go? com is up and running, and they can get a lot of information there. And of course, anyone who wants to speak with us can just click there, and someone will reach out to them very quickly.
Fantastic. We're starting now the fun part of the journey, the real hard work of really starting to go and send a message to people, and we'll be happy to speak with anyone who wants, and really help them become more resilient and be able to revise their applications. com.
D-R-E-E-Z. com. Just so people see it on the bottom third of the screen.
Zeev, I wish you a lot of luck, though I suspect with your background and what you guys are seeing, you'll do just fine, but it's always good to have luck anyway. Do come back, keep us posted on what you're doing and what Zeev's, what Driz is up to. We'd be happy to hear, and fantastic.
Thank you. You're never done in this business. Never done.
They pull you back in. They'll be pulling you in. All good.
Like I said, I hope that's my last backup week. You'll see. All right.
Best of luck. com, here on Textron TV. Check it out.
We'll be back in a minute with more.