Glossary · System coordination, integration and orchestration
Strong consistency
Also known as: Linearizability
German: Starke Konsistenz
In distributed systems, strong consistency is a consistency model in which every read returns the most recent completed write, so all clients see the same data in the same order as if there were a single copy, regardless of which node they access.
- System integration
In one sentence
Strong consistency guarantees every read returns the latest completed write, so all clients see the same data as if there were one copy.
Example
A tool management database uses strong consistency so that two machines can never both check out the same tool at the same moment.
How it applies
- Engineering: Strong consistency needs coordination between nodes (synchronous replication, quorum writes), which increases latency and reduces availability when the network partitions. Use it where conflicting decisions would be costly: resource allocation, inventory, unique assignments.
- Trade-offs: Many systems offer weaker models, such as eventual consistency, for better performance. Choose the model per data type rather than for the whole system.
- Testing: Consistency problems appear under concurrency and failures; test with parallel clients and network interruptions.
- Documentation: State the consistency guarantees of interfaces and data stores in the technical documentation, so integrators do not assume guarantees that are not given.
Strong vs. eventual consistency
With strong consistency, all readers see a write immediately after it completes. With eventual consistency, replicas converge over time, and readers may temporarily see old data; see Replication lag. Eventual consistency is often sufficient for dashboards and reporting, rarely for allocation decisions.