The DevSecOps Talks Podcast

Julien and Andrey got together to define the scale and ways to automate the scaling of your infrastructure in response to changes in load patterns.
What are the prerequisites implementing scaling? What is cooling down, warm up, horizontal and vertical scaling, scale-up, and scale in? What are the metrics that could be useful for making scaling decisions?
And last but not least, the very unexpected spin that Julien gives to the conversation. 

 

Visit https://devsecops.fm to see show notes and https://gitter.im/devsecopstalks/community to join a discussion

This time we are discussing the white paper by Summit Route - AWS Security Maturity Roadmap 2020. Tune in to learn more about the white paper and recommendations that we pile up on top of it.
To view show notes visit https://devsecops.fm
Chat with hosts and suggest topics for upcoming episodes at our Gitter channel https://gitter.im/devsecopstalks/community

Our guest speaker is Anton Babenko
he is DevSecOps Talks podcast fan, AWS Community Hero, Terraform fanatic, HashiCorp Ambassador and a prolific open source contributor.
After listening to episode #9 Terraform in CI and #1 Infrastructure as code,
Anton decided that enough is enough and volunteered to give his point of view on Terragrunt since he though that we are missing a few important points.
In this episode, we are discussing the use cases of Terragrunt,
a wrapper around Terraform for working with multiple environment and modules.

How do you start to implement a CI pipeline when dealing with infrastructure as code implemented via Terraform? What are the security concerns when the credentials to the whole kingdom are used in an automated process? In this episode, we discuss the various security and feasibility aspects of using Terraform in a CI pipeline.

We start the episode by catching up with what we’ve been working on. Feel free to skip to 11:52 if you want to go directly to the topic. Having an automated process to deploy and manage infrastructure has advantages such as fast feedback and collaboration. The code for the infrastructure is treated like an application that is versioned, tested, and deployed.

Show notes are available at https://devsecops.fm/episodes/terraform-in-ci/

Andrey tells us the story of how DevOps came into existence and took over the market. We discuss the marketing around it, its relationship with DevSecOps. We tried to shed a light on what is marketing strategy versus implementing DevOps in an organization. We also compared DevOps to SRE (Site Reliability Engineering)

In this episode, Mattias, Julien, and Andrey share tips and tricks on how to stay on top of what is going on in the industry, resources they use for continuous learning. Make sure to visit devsecops.fm to check out show notes that contain references to resources mentioned during discussion and more

This time Johan Abildskov, a Senior Consultant with Praqma/Eficode, joins us to talk about SemVer (Semantic Versioning), and we finally get to hear what Julien has to say about it. We get to explore different options regarding versioning and how it helps humans communicate. At the end of the podcast, everyone gets to share their approach and recommendations for versioning things.

We had a couple of possible topics for this episode but before getting started with them we decided to discuss what technological problems we were solving during the last two weeks. Well, turns out there was quite a lot to discuss. Tune in for tips on ssh session logging on the ssh server, preventing downloads from AWS S3 even if you got read access, credentials in Git repository 🤦, why you should (or should not) do K8S and more.

---

Summary

In this free-form early episode of DevSecOps Talks, a casual "what have you been up to" catch-up turns into a sharp exchange on the gap between security in theory and security in practice. One host discovers plaintext service account keys, database passwords, and a production SSH tunnel all committed straight into a Git repository — and the team walks through how to unwind that without breaking delivery. Julien Bisconti argues that security tooling is fundamentally failing developers because it is too hard to use under real delivery pressure. The episode also delivers strong opinions on why teams should not default to Kubernetes, the hidden complexity of S3 encryption with KMS keys, and why Google's BeyondCorp model makes VPNs look like a relic.

Key Topics

SSH session logging, bastion hosts, and compliance visibility

The episode opens with a deep dive into SSH session logging for bastion hosts in AWS. One of the hosts explains how AWS Systems Manager Session Manager can be used to access instances without VPNs or direct inbound connectivity — the SSM agent on each instance calls home to AWS, and AWS proxies the connection back. That model is attractive for hybrid and on-prem environments because it removes networking complexity around NAT, port forwarding, and VPN setup. It also provides session logging, IAM-based access control, and command output recording.

But the drawbacks surface quickly. Session Manager logs users in as a generic SSM agent user with /usr/bin as the working directory. Documentation is sparse, and Bash is launched in shell mode to support color interpretation, which pollutes session logs with escape characters. A bigger concern is that access control rests entirely on IAM credentials — in an environment with fully dynamic, short-lived credentials that is manageable, but it becomes risky anywhere static keys exist.

The host describes trying to map Session Manager logins to individual users, only to find that it requires static IAM identities with specially named tags containing usernames — a non-starter for environments where everything is dynamic.

That leads into alternative approaches. An AWS blog post describes forcing SSH connections through the Unix script utility to record sessions, then uploading logs to S3. But even that is fragile: logs are owned by the user, so technically the user can delete or overwrite them. A more robust path is tlog, a terminal I/O logger that writes session data in JSON format to the systemd journal, where it cannot be easily tampered with. From there, the CloudWatch agent can export journal data to S3 for long-term storage.

The broader point is that command logging sounds simple in compliance conversations, but in practice it becomes a deep rabbit hole full of bypasses, noise, and design tradeoffs.

Monitoring user activity without drowning in logs

The hosts compare notes on monitoring shell activity. One host mentions using auditd to track user actions on bastion hosts in a previous environment, but the log volume was overwhelming — even Elasticsearch struggled to keep up with the ingestion rate.

That sparks a discussion around anomaly detection and heuristics. The real challenge is not collecting logs but determining what is unusual and worth investigating. Failed SSH login alerts are mentioned as a useful signal, though another host pushes back: "Should you have SSH with the password at all? You should have a key." The point stands — without careful tuning, even sensible alerts generate noise faster than teams can act on them.

The exchange captures a recurring DevSecOps reality: collecting telemetry is the easy part; turning it into something actionable is where most teams get stuck.

S3 bucket security, public access controls, and KMS encryption surprises

The conversation shifts to AWS S3 security. Public buckets remain a common source of breaches, but AWS now offers S3 Block Public Access — account- and bucket-level settings that prevent public access regardless of individual object ACLs. In Terraform, this is a dedicated resource block.

The more nuanced insight is about encryption. The host explains the difference between S3 server-side encryption with the default AWS-managed key (SSE-S3) and encryption with a customer-managed KMS key (SSE-KMS). With SSE-S3, S3 decrypts objects transparently for any client with read access to the bucket. With a customer-managed KMS key, S3 cannot decrypt the object unless the requester also has kms:Decrypt permission on that specific key.

This became a real problem in a cross-account, cross-region workflow involving Go Lambda binaries. Go Lambdas require the deployment artifact to reside in the same region as the function. The team was copying artifacts between accounts and regions, had granted S3 read permissions, but downloads kept failing. CloudTrail logs revealed the real culprit: "I cannot decrypt." The consumers lacked KMS key access. In that case, the fix was switching to SSE-S3 since the artifacts did not require the stronger protection of a customer-managed key.

The host is careful to note that AWS documentation on cross-account S3 access does not prominently flag this encryption interaction — a gap that can cost teams hours of debugging.

Plaintext secrets in Git: a frighteningly common anti-pattern

One of the most memorable segments comes when a host describes reviewing an application stack and finding service account keys committed in cleartext in the repository root. The repository also contained a large configuration file with usernames, passwords, API credentials for mail services, login providers, and multiple environments (dev, prod) — all in plain text.

But the worst part: for local development, the team SSH-tunneled into the production SQL server, mapping remote port 3306 to local port 3307. An SSH key providing direct access to the production database was sitting right there in the repo.

The reaction is immediate — this is exactly the kind of setup that accumulates when convenience wins over security for too long. But rather than proposing a risky teardown, the host outlines an incremental migration plan:

1. Dockerize the local development stack so developers can run everything locally without production keys
2. Move secrets into the CI/CD pipeline, injecting them only at build and deploy time
3. Separate environments so test clusters get test credentials and production secrets never touch developer machines

Andrey pushes the thinking further: injecting secrets at build time is still risky because anyone who gets the Docker image gets the secrets. The better model is runtime secret retrieval — workloads authenticate dynamically at startup and fetch only the secrets they need. HashiCorp Vault is the concrete example: in a Kubernetes environment, a pod uses its Kubernetes service account to authenticate to Vault, obtains a short-lived token, and retrieves static or dynamic secrets. If someone steals the image and runs it outside the cluster, they cannot authenticate and get nothing.

Vault versus cloud-native secret management

The secrets discussion expands into a broader comparison. Andrey, who has been doing public speaking about Vault and fielding consulting requests around it, frames the choice pragmatically.

For hybrid-cloud or multi-cloud environments, Vault is likely the best option because it provides a unified interface for secret management, dynamic credentials, and synchronization across providers.

For single-cloud commitments — say, all-in on AWS — native services can cover many of the same use cases: AWS STS for temporary credentials, RDS IAM authentication for database logins, AWS Secrets Manager (which may even be running Vault underneath, as one host speculates), and AWS Certificate Manager for TLS certificates. If the organization is not going multi-cloud, the overhead of running Vault may not be justified.

The recommendation is not ideological. It depends on architecture, portability needs, and operational complexity.

When Vault works technically but fails organizationally

Julien Bisconti adds an important caveat from experience. He describes deploying Vault in a multi-availability-zone setup with full redundancy — technically solid. But the project "went to a halt completely" when it hit governance questions: who should access what, under which rules, and who owns the policies. It became a political war, and the entire deployment had to be rolled back.

The lesson: security tools are good at automating technical workflows, but if the underlying organizational process is broken, you automate a broken process. Security, monitoring, deployment, and access control are deeply entangled, and tooling alone cannot untangle them.

Security tooling fails because developers cannot use it

Julien brings the strongest developer-empathy argument of the episode. Developers do not ignore security because they are careless — they bypass it because secure workflows are too awkward under delivery pressure. A manager does not understand why the developer is blocked, pressure mounts, and the result is // just hardcode that here, I don't care, it works.

Even simple tasks illustrate the problem. Julien asks: can you generate an SSL certificate with OpenSSL from memory right now? Most engineers cannot — it is something they do every few months and have to look up each time. He references the famous XKCD comic about entering the correct tar command with ten seconds left.

This evolves into a philosophical observation. One host identifies as a "tool builder" rather than a "product builder" — someone who enjoys building mechanisms but does not always think deeply about end-user experience. That mindset, common among infrastructure and security engineers, may explain why so many DevSecOps tools are powerful but painful to adopt. The gap is not in capability but in usability.

VPNs, zero trust, and the BeyondCorp model

Julien argues that VPNs are an increasingly painful abstraction. Even Cisco — the company that essentially built enterprise VPN technology — had to raise capacity limits during the COVID-19 pandemic because their own infrastructure could not handle the load. Split tunneling introduces its own vulnerabilities, and full-tunnel VPN creates a bottleneck for everything.

He points to Google's BeyondCorp model, published in 2014, which established the principle that network location should not determine access. The analogy: do you build a castle with walls where anyone inside has full access, or do you put a guard in every room checking credentials? The latter — zero trust — is harder to implement, but it limits blast radius and removes the binary "in or out" problem.

Andrey connects this to the emerging service mesh ecosystem. Technologies like Consul Connect implement zero-trust networking at the application level with mutual TLS and identity-based authorization. The hosts note that the service mesh space is still fragmented — just as there was a "war of orchestrators" before Kubernetes emerged as the default, there is now a "war of service meshes" still playing out.

Kubernetes hype versus simpler orchestration

A significant portion of the episode is a productive debate about orchestration choices. Andrey argues strongly against defaulting to Kubernetes. He describes a hybrid-cloud project in Africa running the full HashiCorp stack: Consul for service discovery, configuration, and networking; Nomad for workload scheduling. A team member with relatively little experience got the stack up and running in days.

Andrey outlines the operational weight of Kubernetes: cluster version upgrades where in-place upgrades may skip new security defaults (making full cluster recreation the recommended path), autoscaler configuration layers (pod autoscaler, cluster autoscaler, resource limits), ingress management, YAML sprawl from Helm charts, and a platform that evolves so rapidly it demands continuous learning. He especially warns against running databases in Kubernetes — the statefulness adds pain.

For single-cloud AWS, he argues that ECS is often the better choice: the control plane is free (or nearly so), the per-node overhead is minimal compared to Kubernetes, and AWS handles the operational burden.

Mattias pushes back with a practical counterpoint. Kubernetes provides a consistent platform for diverse workloads — containers, databases, monitoring, custom jobs — all managed through the same interface. Helm charts for common components like nginx-ingress, cert-manager, and external-dns make the ecosystem approachable. The value is in standardization and adaptability.

The hosts also note GKE's pricing evolution: Google introduced a per-cluster management fee (roughly $0.10/hour per control plane) to discourage sprawl and encourage consolidation — a signal that even managed Kubernetes has real costs.

The disagreement is honest but constructive. The shared conclusion: start with what the business needs, then pick the simplest tool that gets you there. "The best battle is the battle you don't fight." And as Julien notes, teams that avoid the Kubernetes default often demonstrate deeper architectural thinking — choosing based on the hype is an insurance policy, but it is not the same as choosing based on needs.

Slack bots, workflow automation, and the security surface

Near the end, Mattias raises the topic of Slack bots for operational tasks — deployment reporting, status checks, and interactive queries. Andrey reframes the conversation around security: if Slack becomes part of a privileged control plane — for example, a bot that handles privilege escalation by requesting approvals through Slack messages — then request spoofing, account compromise, and weak isolation become serious concerns.

The idea of a privilege-escalation bot is interesting (request access via Slack, get approval from designated approvers, receive time-limited credentials with full audit logging), but the attack surface is real. Slack provides a powerful collaboration platform for building workflows without custom UIs, but once it handles access decisions, security design matters as much as convenience.

Highlights

"All the service account keys were in clear text. In the repo."

A host describes opening up a client's application stack and finding cloud service keys, usernames, passwords, API credentials, and an SSH key that tunnels directly into the production SQL server — all committed to Git in plain text. It is the kind of discovery that instantly explains years of hidden risk.

How do you unwind that without breaking delivery? The hosts walk through an incremental migration plan in this episode of DevSecOps Talks.

"Security tooling is actually not that usable."

Julien Bisconti delivers a sharp truth: developers do not bypass security because they are careless. They do it because secure workflows are too slow, too confusing, and too far removed from how they actually work. When the pressure comes from a manager who does not understand the blocker, the shortcut wins every time.

A candid take on why hardcoded secrets keep showing up in real codebases. Listen to the full discussion on DevSecOps Talks.

"I really applaud people who don't choose Kubernetes — that means they actually know what they're doing."

One of the spicier platform takes of the episode. The argument is not that Kubernetes is bad, but that defaulting to it without analyzing your actual needs is a sign of hype-driven architecture. If a simpler stack solves the problem, picking the biggest platform just creates more operational burden.

Hear the full Kubernetes-versus-Nomad-versus-ECS debate on DevSecOps Talks.

"If your process is not good, you're going to automate a bad process."

Julien recounts deploying Vault with full HA and multi-AZ redundancy, only to have the project grind to a halt over organizational politics — who should access what, and who decides. The tooling worked perfectly. The organization did not.

A reminder that DevSecOps maturity is not just about picking better tools. Catch the full story on DevSecOps Talks.

"Once somebody is inside, they have the keys to the kingdom."

The VPN and zero-trust discussion delivers one of the strongest security arguments of the episode. Julien explains why broad network access — the castle-and-moat model — is the wrong abstraction for modern systems, and why identity-based, fine-grained access control is worth the implementation cost.

If the old perimeter model still shapes how your team thinks about infrastructure security, this part of the episode will resonate. Listen on DevSecOps Talks.

Resources

• AWS Systems Manager Session Manager — AWS documentation for Session Manager, which provides secure instance access without SSH keys, open ports, or bastion hosts, with built-in session logging.

• tlog — Terminal I/O Logger — Open-source terminal session recording tool that logs to systemd journal in JSON format, making sessions searchable and tamper-resistant. Discussed in the episode as a more robust alternative to the Unix script command.

• AWS S3 Block Public Access — AWS documentation on account- and bucket-level settings to prevent public access to S3 resources, regardless of individual object ACLs or bucket policies.

• Troubleshooting Cross-Account Access to KMS-Encrypted S3 Buckets — AWS guidance on the exact issue discussed in the episode: S3 downloads failing because the requester lacks KMS key permissions, even when bucket-level access is granted.

• BeyondCorp: A New Approach to Enterprise Security — Google's foundational 2014 paper on zero-trust networking, which established the principle that network location should not determine access. Referenced by Julien in the VPN discussion.

• HashiCorp Nomad — A lightweight workload orchestrator with native Consul and Vault integrations. Discussed as a simpler alternative to Kubernetes, especially for hybrid-cloud and small-team environments.

• Consul Service Mesh (Consul Connect) — HashiCorp's service mesh solution providing zero-trust networking through mutual TLS and identity-based authorization. Mentioned as the networking layer in the Africa hybrid-cloud project.

• XKCD 1168: tar — The comic Julien references about the impossibility of remembering command-line flags — a humorous illustration of why security tooling needs better usability.

In this episode Mattias is trying to convince that running docker in k8s is more security then VM. Did he success ? listen and find out.

---

Summary

Mattias makes a bold claim: Docker containers are more secure than virtual machines. Andrey and Julien push back hard — and by the end, the three hosts explicitly agree to disagree. Along the way, they dig into why container breakouts are harder than people assume, how Lambda micro VMs can be exploited through warm TMP folders, why "containers do not contain" without extra kernel controls, and whether good monitoring matters more for security than any isolation technology. Recorded during COVID-19 lockdowns in 2020, the debate captures a moment when the container-vs-VM argument was far from settled.

Key Topics

Docker vs. VM security: technology vs. ways of working

Mattias opens the main debate by arguing that Docker containers are more secure than VMs in practice. His reasoning: containers are smaller, more focused, and more ephemeral than traditional virtual machines, which reduces attack surface. In a typical VM, you find mail agents, host-based intrusion detection, syslog, monitoring tools, and other services all coexisting with the application. In a container, you ideally run only the application itself.

Andrey pushes back immediately. He argues Mattias is comparing operational models, not technology. A well-run VM can also be immutable and minimal — you redeploy from a new image the same way you replace a container. Likewise, a badly built container can be long-lived, bloated, and full of unnecessary tools. Andrey has seen enterprises that run containers for months, SSH into them, and treat them like VMs.

Mattias concedes the point but maintains that the standard approach differs: VMs are typically kept running longer with more tools, while the standard approach for containers in Kubernetes is to rotate them and keep a smaller footprint. Andrey counters that most Docker images run as root by default, giving attackers more privilege than they would have on a typical VM where processes run under limited service accounts. This is one of the sharpest exchanges in the episode — better tooling does not fix insecure defaults.

The hosts eventually agree that both technologies can be secured well, but do not reach consensus on which is easier. Andrey summarizes it cleanly: containers make it "a little bit easier" to do the right thing because they narrow the focus to the application rather than the entire operating system, but it is absolutely possible to reach the same security level with VMs.

Why container breakout is not as trivial as people imply

Mattias challenges the common assumption that containers are unsafe because "you can break out of them." He points out that every container breakout CVE he has reviewed requires significant preconditions: either running an attacker-controlled image or running in privileged mode. You cannot take a standard Ubuntu container image, run a single command, and escape. The threat is real but requires chained attacks, not a single exploit.

Julien and Andrey accept the premise but note that the comparison matters. VM isolation is fundamentally stronger at the hypervisor level. Container breakout may be hard, but it is architecturally easier than VM escape. The discussion reframes the question: runtime security is less about one isolation boundary and more about how many obstacles an attacker must pass through.

Micro VMs, Firecracker, and Lambda attack vectors

Andrey brings up an important middle ground between containers and VMs: micro VMs. AWS Lambda runs on Firecracker, an open-source micro VM monitor. Lambdas are ephemeral, have read-only file systems, minimal tooling, and no access to source code or settings — making them quite secure by design.

But Andrey describes a real attack path researchers have demonstrated. The /tmp directory in Lambda is writable. If an attacker exploits a vulnerability to get code execution within the Lambda, and the Lambda is kept warm (invoked within 15 minutes so it stays in memory), the /tmp folder persists between invocations. An attacker can download tools incrementally across multiple Lambda runs, building up capability over time. From there, they can explore IAM permissions, exfiltrate data by encoding it in resource tags, or even override the Lambda function itself.

The point is that even well-designed ephemeral environments have attack paths when defenders are not paying attention. Security depends on hardening and monitoring, not just on the isolation primitive.

Containers do not contain: AppArmor, Seccomp, and policy controls

Julien delivers the episode's sharpest technical point: "Containers do not contain." They are primarily Linux namespace isolation and need additional kernel controls — AppArmor profiles and Seccomp filters — to properly restrict what applications can do at runtime. Without those extra layers, a container running as root is effectively root on the host machine, and a container with host network access is the same as running directly on the server.

This shifts security responsibility in uncomfortable ways. In VM environments, operations and security teams traditionally handle access controls. In containerized environments, developers are often expected to define security profiles for their workloads — but they may not know which system calls or privileges their applications need. Julien describes this as a fundamental organizational gap: the people writing the workload and the people securing the workload are rarely working hand in hand.

Mattias suggests that platform teams can solve this by enforcing policies centrally. He references tools like Open Policy Agent to set standards for what gets deployed into a cluster, rather than relying on every developer to configure security correctly.

Kubernetes makes monitoring and response easier

Mattias makes a strong case for container platforms as detection and response environments. He describes working with Falco, a runtime security tool, and highlights a powerful capability: if someone opens a shell inside a container, Falco can detect that behavior and the container is killed automatically. That kind of automated response is natural in an environment built around disposable workloads. On a VM, shells are a normal part of operations, making the same detection much harder to act on.

Julien extends this into a broader argument about monitoring and security being inseparable. He argues that when monitoring is poor, access control becomes chaotic — developers need broad production access just to debug issues. But with strong observability, teams can use feature flags, targeted routing, and centralized logging instead of SSH-ing into production. Good monitoring reduces the need for risky access patterns.

Julien offers a practical example: instead of blocking developers from opening shells in containers, observe that they are doing it and ask why. If they need logs, build a secure log access API. If they need to debug, improve the observability tooling. Monitoring turns security violations into product requirements.

Minimizing container images

Julien mentions using DockerSlim (now SlimToolkit) to strip unnecessary components from container images, reducing attack surface without requiring deep knowledge of every dependency. It is not a complete security solution, but it is an easy first step that removes much of the bloat containers inherit from their base images.

For organizations with compliance requirements, Julien notes that third-party security vendors provide validated runtime solutions — useful for audit purposes where you need a third party to confirm that the running workload matches what was built internally.

Bundling dependencies with the application

Mattias raises a concern about how containerization changes dependency management. In older models, operations maintained the web server (Apache, Nginx) separately from the application. In containers, the web server, runtime, and application are bundled together. That means patching the web server requires rebuilding and redeploying the entire container, even when the application code has not changed.

Andrey reframes this as a different packaging model, not a new problem. With Java WAR files deployed to Tomcat, you already had dependency coupling — you just managed it differently. Containers actually improve the situation in one way: each application owns its own dependency lifecycle instead of sharing an application server. One application can upgrade independently without affecting others on the same host.

Both hosts note that dedicated application servers are fading. Modern applications in Go, Python, and Node.js often handle HTTP directly, removing the need for a separate web server entirely. The ingress controller in Kubernetes handles routing at the cluster level, which is a separate concern from the application.

The hosts agree to disagree

The episode ends without consensus. Mattias remains firmly convinced that containers, run properly in Kubernetes, are more secure than VMs. Julien's final position: "Containers can be as secure as VMs, but they need more work to get there." Andrey advocates for a layered approach — use both VMs and containers, with container security focused on application concerns and VM security focused on operational and resource isolation. He also notes that CoreOS, once the go-to minimal container OS, had recently been discontinued by IBM, leaving teams to find alternatives like Fedora CoreOS.

Highlights

"Containers do not contain."

Julien delivers the episode's most quotable line, reminding listeners that containers are mostly Linux namespacing — not real isolation boundaries. Without AppArmor, Seccomp, and careful configuration, a container is far less restrictive than people assume. A sharp reality check for anyone treating "containerized" as synonymous with "secure." Listen to the full episode on DevSecOps Talks to hear why container security is never just about packaging.

"If somebody pops a shell in a container, that container is killed."

Mattias describes working with Falco and highlights a capability that captures the strongest pro-container argument: disposable workloads change the incident response model entirely. On a VM, a shell is normal. In a container, it is an alarm — and the platform can act on it automatically. Listen to the episode to hear how the hosts connect runtime detection, monitoring, and automated response.

"Most of the Docker images out there are running as root."

Just when the debate leans in Docker's favor, Mattias himself brings it crashing back. On VMs, running as root is rare. In containers, it is the default. Better tooling does not fix insecure defaults — and this remains one of the most practical risks in container environments. Hear the full back-and-forth on DevSecOps Talks.

"We have to separate apples from bananas — the technology from the ways of working."

Mattias draws a sharp line that reframes the entire debate. Are containers actually more secure, or are teams comparing modern container practices against outdated VM operations? A useful reminder that architecture arguments often hide workflow arguments underneath. Listen to the full conversation for the spirited disagreement that follows.

"Monitoring very much goes hand in hand with security."

Julien makes the case that bad observability leads directly to bad access control. When developers cannot see what is happening in production safely, they need more privileges, more access, and more risky workarounds. Fix the monitoring, and many security problems solve themselves. Listen to the episode on DevSecOps Talks to hear why observability might be the most underrated security control.

"Containers can be as secure as VMs, but they need more work."

Julien's final verdict — delivered over Mattias's loud objections — perfectly captures the episode's unresolved tension. The hosts explicitly agree to disagree, making this one of the more honest security debates you will hear on a podcast. Catch the full exchange on DevSecOps Talks.

Resources

• Falco — CNCF-graduated runtime security tool that detects anomalous behavior in containers and Kubernetes using eBPF. Mentioned by Mattias for its ability to automatically kill containers when suspicious activity like shell access is detected.

• Firecracker — Open-source micro VM monitor built by AWS, powering Lambda and Fargate. Discussed by Andrey as an example of ephemeral, hardened execution environments and their attack surfaces.

• SlimToolkit (formerly DockerSlim) — Tool for analyzing and minimizing container images, automatically generating AppArmor and Seccomp profiles. Mentioned by Julien as a practical way to reduce attack surface without deep security expertise.

• Open Policy Agent (OPA) — General-purpose policy engine for enforcing security and operational policies across Kubernetes clusters. Referenced by Mattias for centrally enforcing deployment standards.

• AppArmor — Linux kernel security module that restricts application capabilities through mandatory access control profiles. Discussed by Julien as an essential add-on for meaningful container isolation.

• Seccomp (Secure Computing Mode) — Linux kernel facility that restricts which system calls a process can make. Used by Docker and Kubernetes to reduce the container attack surface by blocking unnecessary syscalls.

• Fedora CoreOS — Successor to CoreOS Container Linux (discontinued 2020), a minimal, auto-updating operating system designed for running containerized workloads. Relevant context for Andrey's mention of CoreOS being killed by IBM.

Your docker images and build are be coming the base for our platform. But are they secure? In this episode we talk about how you can secure your docker images.

---

Summary

In this early DevSecOps Talks episode, Mattias, Andrey, and Julien dig into Docker security as a supply chain problem — and quickly dismantle the assumption that a signed container means you know what is inside. Julien pushes back sharply: signing only gives a "semantic guarantee" that an image is what it claims to be, not that it is safe. Mattias argues that containers were designed to be convenient, not secure by default, while Andrey points out that containerization has fundamentally changed the patching game — once the OS, web server, and application are packaged together, every security fix becomes a rebuild-and-redeploy exercise. The hosts make the case for layered scanning, slim runtime images, multi-stage builds, and continuous rebuilding as the practical path to running containers safely in production.

Key Topics

Container images vs. running containers

The conversation starts by separating two distinct parts of container security: the image and the running container.

Mattias explains that a container image can be treated much like any other file or archive — a zip or tar file sitting on disk. Because of that, teams can sign images cryptographically to verify origin and integrity, similar to how Node.js developers sign releases with their private keys. That gives consumers confidence that the image came from a known source and has not been tampered with.

But Julien pushes back on a common misunderstanding: signing does not mean the contents are inherently safe. As he puts it, you get a "semantic guarantee that this image is what it's pretending to be" — but not proof that everything inside is secure. Authenticity is not the same as security.

The hosts frame this as a trust problem. In a production cluster, teams often want to prevent engineers or workloads from pulling arbitrary images and running them without controls. Signed images and curated registries help, but they do not eliminate the need for careful validation.

Trust, Docker Hub, and the container supply chain

A major part of the episode focuses on how much trust teams should place in public images, including those from Docker Hub.

Andrey raises the practical reality: if you are running four different languages, you cannot build and maintain base images for all of them. It is much easier to grab the latest Node.js, Python, Ruby, or Java images from Docker Hub and build from there. Julien and Mattias acknowledge that reality, but caution against treating "official" or branded images as automatically secure.

Julien walks through the different trust levels on Docker Hub:

• Images from unknown individuals are the hardest to trust
• Organization-backed images (Red Hat, CloudBees, etc.) provide more accountability based on brand recognition
• Even reputable images can contain known vulnerabilities — scanning a Jenkins image from Docker Hub can reveal a surprising number of CVEs
• A trusted source can still introduce problems, whether by mistake or through malicious intent

That leads into a broader discussion of supply chain attacks. Julien references real examples where Node.js libraries on npm were taken over by malicious parties after the original maintainer walked away. The same risk applies to container images.

Julien points out that large organizations sometimes go as far as rebuilding all dependencies from source — he mentions having heard of teams that do not pull jar files from Maven Central but build their own from source to verify exactly what they are shipping. While that is not feasible for every team, the principle stands: reduce blind trust and increase verification where the environment demands it.

Why container security is not just image signing

The discussion then shifts from image authenticity to runtime security.

Mattias explains that containers rely on Linux kernel primitives — namespaces for process isolation, along with controls for networking, memory, and disk. These low-level APIs are useful for resource sharing and scaling, but they were not originally designed as strong security boundaries. As he puts it, "the container does not contain things, it's just an abstraction." Container breakout vulnerabilities matter because an attacker who can exploit the runtime or host interface may reach beyond the container itself.

This leads to one of the episode's sharpest observations from Mattias: containers became popular because they are efficient and convenient to operate — you can bin-pack them on the same hardware and run far more applications per server. But from a security perspective, "it was not designed to be secure by default, it was designed to be convenient." That gap between convenience and security is what teams must actively address through scanning, hardening, and runtime controls.

CVE scanning: registries, dependencies, and source code

The hosts spend a good amount of time discussing scanning tools and where each fits in the security pipeline.

Mattias notes that most container registries now offer built-in vulnerability scanning, sometimes called container analysis APIs. Julien suggests a practical AWS-based pattern: if you do not want to pay for Docker Hub premium but still want to use public images, you can pull from Docker Hub, push into AWS Elastic Container Registry (ECR), and take advantage of its built-in CVE scanning. Then you restrict your production orchestrators to pull only from ECR.

Julien draws an important distinction between types of scanning:

• Registry scanning examines what packages are installed in the image at the OS level. The registry unpacks the image, identifies installed packages, and flags known CVEs.
• Dependency scanning tools like Snyk, Dependabot, and similar platforms check application dependency manifests (package.json, requirements.txt, pom.xml) against CVE databases. They are protecting against supply chain vulnerabilities in libraries, not scanning custom application code.
• Static analysis and linters can catch some obvious security issues in application source code, but as Andrey notes, they mainly catch "easy targets" and default patterns.

Julien initially states that registry scans do not cover source code, then corrects himself to clarify the distinction more precisely: registries scan installed OS packages, while separate tools scan programming language dependencies. Neither deeply analyzes your own custom code. That leaves an unknown component in the stack that teams need to address through other means — code review, testing, and secure development practices.

Andrey also mentions using Anchore, which he describes as the foundation for many of these CVE scanning capabilities.

The shift from OS patching to image rebuilding

One of the most practical insights comes from Andrey, who compares containers to older operational models.

In traditional environments, teams could patch the operating system or update components like Nginx independently of the application. With containers, those layers are packaged together. If a new Nginx vulnerability is disclosed, the team needs to rebuild and redeploy the entire image that contains both the web server and the application code.

This changes patching from an infrastructure task into an application delivery task. Security updates are no longer something ops handles in isolation — they flow through the same build-and-deploy pipeline as feature code.

The hosts argue that this is why security must be a concern from the earliest stages. As Andrey puts it, referencing Julien's earlier point: security belongs in the first commit, because that is when it is cheapest and easiest to get right. A green build today does not guarantee a safe deployment tomorrow if new CVEs are published against the packages already running in production.

Slim images, distroless approaches, and DockerSlim

Mattias argues strongly for reducing container contents to the bare minimum. He highlights DockerSlim (now SlimToolkit), a project he uses frequently that strips images down to only the components essential for the application. In his example, a Maven-based application image dropped from roughly 600 MB to 140 MB — with no bash shell or other standard OS tooling left in the result.

Julien reinforces the security rationale: "the less code you have, the less vulnerability you have, and that's what you want in production." He mentions Alpine Linux and Google's distroless images as complementary approaches that aim for the same goal — minimal OS footprint in production containers.

The common theme is that production containers should not carry build tools, shells, package managers, or debugging utilities. Every unnecessary binary is a potential attack surface. The best production image is not the one easiest to build, but the one that contains the least unnecessary code.

Multi-stage builds and separate build vs. runtime images

The hosts spend considerable time on one of the most practical Docker security patterns: multi-stage builds.

Julien explains the concept of build stages — using an intermediate container with all build dependencies to compile the application, then copying only the final artifact into a much smaller production image. This separation means the production image does not need compilers, package managers, or the full dependency tree.

Andrey confirms this maps directly to Docker's multi-stage build feature: "You just build your Docker build in one stage and then just copy build results to the next stage." He also points out the developer experience benefit — since the build environment is defined inside the Dockerfile itself, developers do not need to set up different language toolchains on their local machines when working across multiple microservices.

Julien adds a performance angle: pulling a pre-built container image with cached dependencies is often much faster than resolving and fetching all dependencies from scratch. He has seen Maven builds that took 20 minutes purely because they had to re-fetch all artifacts every time. Pre-building and caching the dependency layer can dramatically improve total build-to-production time.

Continuous rebuilding and reducing attacker persistence

Andrey recommends reducing the lifetime of deployed images by rebuilding base images and all derived containers regularly — potentially every week — pulling in the latest patches each time. While this adds operational overhead, it shortens the window of exposure and makes it significantly harder for attackers to maintain persistence in stale environments.

Julien frames this as a recurring maintenance budget that every engineering team must accept. As he puts it, "if you don't spend at least one day per week updating the stuff, it's going to accumulate over a year or something. And then you have to spend two weeks fixing all that." The compound interest on security debt is steep.

Tags, digests, signing, and private registries

Near the end of the episode, Mattias raises a practical deployment question: how should teams store and reference images securely? He contrasts mutable tags (which can be overwritten on Docker Hub) with immutable SHA-based digests, image signing, and private registries — and admits there are so many options it is hard to know where to start.

Julien recommends implementing all of these controls, but not all at once. He advocates for an incremental approach: define your security objectives, then build toward them layer by layer. Start with what gives the most immediate protection and expand from there.

The hosts do not present a single silver bullet. Instead, they emphasize defense in depth: scanning at every level (code dependencies, container base images, production images), signing for authenticity, private registries for access control, and infrastructure-level enforcement.

Build pipeline security and handling secrets

The episode closes by touching on a problem the hosts agree deserves its own dedicated discussion: securing the build system itself.

Mattias points out that the build server has access to source code, credentials, signing keys, registries, and deployment systems. If an attacker compromises it, they can inject malicious code during the build process — effectively poisoning everything downstream.

The hosts then discuss the challenge of passing credentials into container builds for private dependencies. Andrey notes that recent Docker versions support passing SSH agents and secrets more safely during builds. He recommends using short-lived credentials (like AWS STS tokens with 15-minute expiration) so that even if credentials leak into image layers, they are already expired by the time anyone could exploit them. He also mentions using IMG, a daemonless image builder, as an alternative to Docker that avoids the need for a Docker daemon during builds.

Julien takes a different approach to runtime secrets: encrypting them with KMS and storing them in a cloud bucket, then fetching them only at container startup. He observes that the real cloud vendor lock-in is never the runtime — "it's always the IAM" — because authorization and access control mechanisms are deeply cloud-specific and difficult to migrate.

Julien adds that handling build secrets often becomes an awkward "dance" of fetching credentials, granting temporary access, and cleaning up afterward. It works, but it remains operationally clumsy.

The hosts agree that build server hardening and the connection between security and cost management (which Julien briefly mentions as natural partners, since understanding who has access to what benefits both) are topics worthy of their own future episodes.

Highlights

"You don't know what's inside — you only have a semantic guarantee."

Julien cuts through a common assumption in container security: signing an image proves origin, not safety. That distinction shapes the entire episode, as the hosts explore why authenticity, trust, and actual security are three separate problems. Listen to this episode of DevSecOps Talks for a grounded discussion on what image signing can — and cannot — guarantee.

"Containers were designed to be convenient, not secure by default."

Mattias makes one of the sharpest points of the episode: containers became popular because they are efficient and easy to operate, not because they provide strong isolation. The container "does not contain things, it's just an abstraction." That is why runtime hardening and vulnerability management still matter so much. Listen to DevSecOps Talks to hear why container adoption created as many security questions as it solved.

"Official on Docker Hub doesn't mean secure — scan a Jenkins image and you'd be surprised."

Julien challenges the idea that a branded or official image should be trusted blindly. Even well-known organization-backed images can contain a surprising number of CVEs, and reputable sources can still introduce malicious changes — intentionally or by mistake. Listen to this DevSecOps Talks episode for a practical conversation about defining trust in your container supply chain.

"The less code you have, the less vulnerability you have."

Julien sums up a recurring theme: smaller runtime images are not just cleaner — they are fundamentally safer. From DockerSlim shrinking a 600 MB Maven image to 140 MB, to Alpine and distroless approaches, the hosts argue for removing everything production does not absolutely need. Listen to DevSecOps Talks to hear why image size and security are more connected than many teams realize.

"Nginx gets a CVE? Now you have to rebuild your entire app."

Andrey highlights how containerization merged the OS patching cycle with the application delivery cycle. In the old world, ops could patch Nginx without touching the app. In the container world, every security update means a full image rebuild and redeploy — making security an application delivery concern, not just an infrastructure one. Listen to this DevSecOps Talks episode for a practical take on why modern patching must flow through the CI/CD pipeline.

"If you don't spend one day a week updating, you'll spend two weeks fixing it later."

Julien describes dependency and image maintenance as a non-negotiable recurring budget. Skip the updates and the security debt compounds fast — turning routine maintenance into an emergency remediation project. Listen to DevSecOps Talks for an honest take on the operational cost of staying secure in containerized environments.

"The real lock-in is never the runtime — it's always the IAM."

In a brief but pointed aside about handling secrets in containers, Julien observes that authorization and access control are the truly cloud-specific parts of any architecture. Runtime workloads can move; IAM policies cannot. Listen to this DevSecOps Talks episode for a candid discussion on where the real complexity lies in cloud-native security.

Resources

• SlimToolkit (formerly DockerSlim) — Open-source tool that minifies container images by removing non-essential components, reducing image size and attack surface without code changes. Mentioned by Mattias in the episode.

• Google Distroless Container Images — Minimal container base images from Google that contain only the application and its runtime dependencies, stripping out shells, package managers, and OS utilities.

• Docker Multi-Stage Builds — Official Docker documentation on using multiple build stages to produce smaller, cleaner production images by separating the build environment from the runtime image.

• Docker Content Trust — Docker's built-in mechanism for cryptographic signing and verification of image integrity and publisher identity using Notary.

• Amazon ECR Image Scanning — AWS documentation on scanning container images for OS and language package vulnerabilities in Elastic Container Registry, mentioned by Julien as a practical alternative to paid Docker Hub scanning.

• Snyk Container — Developer security tool for scanning container images and application dependencies for known vulnerabilities, with remediation guidance and base image upgrade recommendations.

• Anchore Container Scanning — SBOM-powered container vulnerability scanning platform, referenced by Andrey as the engine behind many registry-level CVE scanning capabilities.

• Alpine Linux Docker Image — Minimal 5 MB base image built on musl libc and BusyBox, widely used as a lightweight, security-conscious alternative to full Linux distribution base images.