Managed NoSQL Database across clouds
Key-value, document and wide-column databases - DynamoDB, Cosmos DB, Firestore, Bigtable, OCI NoSQL, Keyspaces, DocumentDB - with point-in-time recovery on and the endpoint private.
11 verified modules, 3 of them live-tested apply→verify→destroy; the rest are static-validated, live-test pending.
Compare by provider
| Provider | Module | Verification |
|---|---|---|
| Alibaba Cloud | A Tablestore Instance Reachable Only from a VPC, with Encrypted Tables That Expire | static-validated |
| AWS | A Cassandra Table that Can Be Restored | static-validated |
| AWS | DocumentDB Cluster, Encrypted and Auditable | static-validated |
| AWS | DynamoDB Table | ✓ live-tested |
| Azure | Azure Cosmos DB Account | ✓ live-tested |
| Google Cloud | A Database whose Irreversible Choices Are Made Deliberately | static-validated |
| Google Cloud | Cloud Bigtable Instance & Table | ✓ live-tested |
| Huawei Cloud | A MongoDB-Compatible DDS Instance with TLS On and the Flavor List the Mode Requires | static-validated |
| IBM Cloud | A Cloudant Instance, IAM-Only, with CORS Off and Document Reads in the Audit Trail | static-validated |
| Oracle Cloud | A NoSQL Table that Is Not Reclaimed Behind Your Back | static-validated |
| Tencent Cloud | A MongoDB Replica Set in a VPC, Encrypted at Rest, Three Nodes Unless You Say Otherwise | static-validated |
How to choose
Decide the data model before the cloud: a document store, a wide-column store and a key-value table are not interchangeable and the migration between them is a rewrite. Then compare the consistency you get by default, whether point-in-time recovery is on or a checkbox, and how the bill is shaped (provisioned capacity, on-demand, or nodes).
When not to use
These scale by giving up the join. A schema that needs relations belongs in the relational hub; a NoSQL table with five access patterns nobody designed for is where the cost surprises live.