Key-Value Database

What each cloud calls it

Features

• AWS DynamoDB, Azure Cosmos DB, and GCP Bigtable are proprietary distributed databases with no direct equivalent from any European cloud provider. None of the five European providers in this table offer a managed key-value or wide-column database in the DynamoDB or Cosmos DB sense: partitioned, globally distributed, with integrated streams and serverless capacity modes.
• This is one of the clearest capability gaps between US hyperscalers and the current European provider landscape. Teams migrating workloads that depend heavily on DynamoDB or Cosmos DB will need to make an architectural decision, not just a provider swap.
• What European providers do offer is managed Redis (see In-Memory Cache) and managed PostgreSQL (see Managed Relational Database). Neither is a like-for-like replacement for DynamoDB, but they cover a significant portion of the access patterns that DynamoDB is used for in practice.

Migration considerations

• Assess actual access patterns first. DynamoDB is frequently used for workloads that do not require its distinctive features: global tables, streams, or serverless auto-scaling. If your application uses DynamoDB as a simple key-value store with predictable throughput, a managed PostgreSQL table with a primary key index or a managed Redis instance may be a practical replacement with modest application changes.
• Self-hosted alternatives on European infrastructure. ScyllaDB, Apache Cassandra, and CockroachDB can be deployed on European VMs or managed Kubernetes clusters. This gives you a DynamoDB-like data model and throughput characteristics without the proprietary API surface. The trade-off is operational responsibility for the database cluster.
• DynamoDB Streams replacements. If your architecture uses DynamoDB Streams to trigger downstream processing (Lambda functions, analytics pipelines), the replacement options on European infrastructure are PostgreSQL logical replication with a consumer, or a message queue (Scaleway Queues, STACKIT RabbitMQ) fed by application writes.
• DynamoDB's on-demand pricing model (pay per request, no provisioned capacity) has no equivalent in the European landscape. Self-hosted alternatives require right-sizing upfront. Model your expected throughput carefully before selecting an instance size.
• Cosmos DB multi-model workloads. Cosmos DB supports document, graph, and table APIs alongside key-value access. If your workload uses the document API, managed MongoDB on OVHcloud, Scaleway, STACKIT, or IONOS is a closer target (see Document Database).

Further reading

• Why European companies are migrating away from US cloud giants→
• European cloud providers are getting closer to parity. What changed?→