11 Services with Docker and Compose
Modern applications are built from interdependent processes that must be discoverable and reliably available over networks. This chapter introduces “services” as the abstraction for those network-accessible workloads, and shows how Docker’s Swarm mode provides first-class tooling to manage them. An autonomous orchestrator continually reconciles a user’s desired state with the current state—starting, stopping, and replacing containers as needed—so operators can focus on intent rather than low-level mechanics. The concepts and workflows presented map cleanly to other container orchestrators, making them broadly applicable.
Starting with a simple web service, the chapter demonstrates enabling Swarm mode, creating a service, and observing how tasks (one container per task) implement service replicas. Key capabilities include self-healing (automatic resurrection of failed containers), scaling via replicated or global modes, and controlled rollouts that specify update order, batch size, and delays until the system converges. It also covers resilient deployment strategies: pausing on failures, rolling back automatically based on thresholds, and defining service health independently of process status with health checks configurable in images or service definitions, including retry, interval, timeout, and the option to disable checks when inappropriate.
The second half moves from imperative commands to declarative environments using Compose V3 and stacks. It explains the YAML-based model, how docker stack deploy creates and updates collections of services, and how changes like scaling are applied without tearing down the stack, with explicit pruning for safe deletions. For stateful workloads, it models named volumes at the stack level and mounts them in services to persist data across restarts and replacements. Finally, it details Docker’s networking for services: the ingress network for published ports, per-stack overlay networks for service-to-service communication, built-in DNS that resolves service names to virtual IPs, and integrated load balancing across replicas, along with Compose options to create, share, or isolate networks as needed.
Initializing a Swarm node.
Creating your first service.
The Swarm node automatically creates a container to run the service software.
Timeline of Swarm reactions to changes in service specification and state.
Timeline of Swarm automated deployment for updated service definition.
Declarative processing loop
Docker network topology, service virtual IP addresses, and load balancing.
Summary
This chapter introduces higher-level Docker abstractions and working with the Docker declarative tooling, and using Compose to model the desired state for multi-service applications. Compose and declarative tooling relieve much of the tedium associated with command-line management of containers. The chapter covers the following:
- A service is any processes, functionality, or data that must be discoverable and available over a network.
- Orchestrators like the Swarm track and automatically reconcile user-provided desired state and the current state of Docker objects like services, volumes, networks.
- Orchestrators automate service replication, resurrection, deployments, health checking, and rollback.
- The “desired state” is what the user wants the system to be doing, or what it is “supposed” to be doing. People can describe desired state in a declarative style using Compose files.
- Compose files are structured documents represented in YAML.
- Declarative environment descriptions with Compose enables environment versioning, sharing, iteration, and consistency.
- Compose can model services, volumes, and networks. Consult the official Compose file reference material for a full description of its capabilities.
Docker in Action, Second Edition ebook for free
