Great question—this is a super common point of confusion when choosing between container runtimes, especially when security is a top priority. Let’s break down how LXD stacks up against both standard Docker and CIS-hardened Docker, with practical context for your decision.

Core Architectural Security Foundations #

First, it’s critical to understand the fundamental differences in how these tools approach isolation:

• LXD is built on top of LXC, designed for system containers (think lightweight VMs that share the host kernel by default, but can also run as full VMs with their own kernel via QEMU). It prioritizes strong out-of-the-box isolation using kernel features like namespaces, cgroups, AppArmor/SELinux, and user namespace mapping.
• Docker is optimized for application containers—focused on packaging single apps and their dependencies. Its default configuration leans toward ease of use, which means some security controls are disabled or relaxed by default.

LXD vs. Standard Docker #

Let’s compare their security profiles straight out of the box:

• Isolation Strength: LXD’s default isolation is significantly stricter. It enables AppArmor/SELinux policies by default, restricts access to sensitive host devices (like /dev/kmem), and limits system calls more aggressively. Docker, by contrast, uses a more permissive default AppArmor profile and allows broader system call access unless explicitly restricted.
• Privilege Management: LXD automatically maps container-internal root to a non-privileged user on the host (via user namespaces), meaning even if an attacker escapes the container, they won’t have root access on the host. Standard Docker does NOT enable user namespace mapping by default—most containers run as root inside, and if an escape occurs, the attacker gains root privileges on the host.
• Attack Surface: Docker’s daemon (dockerd) runs as root and exposes a wide API surface, which can be a risk if not properly secured. LXD’s daemon also runs as root, but its API is more focused on system container management, with fewer exposed endpoints that could be exploited. Additionally, Docker’s vast ecosystem of third-party images often includes unnecessary privileges or insecure configurations, increasing attack surface.
• Host Access Risks: Docker allows easy mounting of host directories by default, which can lead to accidental or malicious data exposure if misconfigured. LXD restricts host filesystem access by default, requiring explicit permission to mount host paths.

LXD vs. CIS-Hardened Docker #

CIS-hardened Docker applies the Center for Internet Security’s benchmark guidelines to lock down Docker’s configuration—this includes enabling user namespaces, enforcing strict AppArmor/SELinux policies, disabling privileged containers, restricting system calls, and blocking sensitive host mounts. Here’s how it compares to LXD:

• Isolation Parity: When properly hardened, Docker can match LXD’s container-mode isolation strength. However, LXD’s VM mode (which runs a full guest kernel via QEMU) offers hardware-level isolation that no Docker configuration can achieve—this is a game-changer for multi-tenancy or handling highly sensitive data.
• Default vs. Manual Effort: LXD comes with most CIS-aligned security controls enabled by default. With Docker, you have to manually apply (or use tools to automate) CIS hardening steps, which adds ongoing maintenance overhead. If you forget to update your hardening config after a Docker upgrade, you could introduce vulnerabilities.
• Flexibility: LXD lets you adjust isolation levels on a per-container basis—you can run a lightweight system container, a fully isolated VM, or something in between. CIS-hardened Docker is more rigid, as it enforces strict baseline rules that may not fit all use cases (e.g., if you need a container with limited privileged access for a specific tool).
• Attack Surface Reduction: CIS-hardened Docker cuts down Docker’s attack surface by disabling unnecessary features (like Docker Swarm if unused) and restricting API access. LXD’s attack surface is already smaller by design, since it’s not built to support the same breadth of application-focused features as Docker.

Practical Recommendations for Your Switch #

• If maximizing isolation (e.g., multi-tenant environments, sensitive workloads) is your top priority: Go with LXD, specifically its VM mode. It’s more secure than even CIS-hardened Docker.
• If you’re running application-focused workloads and want a mature ecosystem: CIS-hardened Docker can get you to a similar security level as LXD’s container mode, but be prepared to maintain the hardening configuration long-term.
• If you value out-of-the-box security with minimal effort: LXD is the better choice—you won’t have to spend hours configuring and maintaining hardening rules like you would with Docker.

内容的提问来源于stack exchange，提问作者inf3rno