Infrastructure, Systems, and Connectivity
Docker Model Runner lets AI models run directly on host hardware instead of inside containers, giving them fast access to GPUs while staying fully integrated with the Docker toolchain. This article covers how it works, how to pull and test models, and how to connect it to Compose-based chatbot apps.
WebAssembly (Wasm) is emerging as a lightweight alternative to traditional containers, and Docker now supports building, sharing, and running Wasm apps using familiar tools. This article walks through writing a simple Wasm app with Spin, containerizing it with Docker, and running it as a Wasm container.
Docker Swarm turns a group of Docker nodes into a secure, highly available cluster with built-in application orchestration. This article covers building a multi-node swarm, deploying a microservices app declaratively, and performing rolling updates without losing desired-state consistency.
Docker networking is built on an open design called the Container Network Model, implemented through libnetwork and extended by pluggable drivers. This article covers the theory behind Docker networking and walks through creating and testing single-host bridge networks, including name resolution and port mapping.
Beyond basic bridge networks, Docker can connect containers directly to existing physical VLANs, resolve service names automatically, and distribute traffic across a Swarm cluster. This article covers the macvlan driver, Docker's built-in service discovery, and Swarm's ingress load balancing mesh.
Overlay networks let containers on different hosts communicate as if they shared a single flat network, forming the backbone of most cloud-native microservices apps. This article covers how to build, encrypt, and test a Docker overlay network across a Swarm cluster, and explains the VXLAN tunneling technology powering it behind the scenes.
Containers get a temporary writable layer by default, which disappears the moment a container is deleted — fine for scratch data, but risky for anything worth keeping. This article explains how Docker volumes decouple persistent data from container lifecycles, and walks through creating, using, and sharing them safely.
Docker's security model is built on layered defense, combining well-established Linux kernel technologies with sensible defaults out of the box. This article explains how namespaces, control groups, capabilities, mandatory access control, and seccomp work together to isolate and secure containers.
Every program a person writes eventually becomes electrical signals moving through silicon. This article explains what computer architecture actually studies, walks through the eight foundational ideas that shape modern processor design, and traces the journey a program takes from human-readable code down to the hardware that runs it.
Opening up a computer reveals a small set of recurring building blocks that work together regardless of the device's size or purpose. This article walks through the core hardware components every system relies on, explains how those components are physically manufactured from raw silicon, and shows the correct way to measure and compare processor performance.
For decades, processors got faster mainly by running at higher clock speeds. That approach hit a physical limit tied to power consumption and heat, forcing the entire industry to change direction toward multiple processing cores instead. This article explains why the power wall happened, how the industry responded with multiprocessors, what real benchmark numbers reveal about performance claims, and common misconceptions to avoid when reasoning about hardware performance.
Every high-level statement a programmer writes eventually breaks down into a small, rigid set of hardware-level operations. This article explains why instruction sets are kept deliberately simple, walks through the core arithmetic and data-movement operations a processor supports, and explains how operands such as registers and memory locations are represented and accessed at the hardware level.
