Managed Search across clouds
Managed search and analytics engines - OpenSearch on AWS and OCI, OpenSearch Serverless, Azure AI Search - configured so the endpoint is private and the index does not grow forever.
9 verified modules - all static-validated and publish-checked; live-test pending.
Compare by provider
| Provider | Module | Verification |
|---|---|---|
| Alibaba Cloud | An Elasticsearch Cluster with Kibana Off the Internet, Which the Provider Does Not Default To | static-validated |
| AWS | A Serverless Collection Reachable from Your VPC with Access Granted by Name | static-validated |
| AWS | OpenSearch Domain that Refuses to Be Public | static-validated |
| Azure | A Search Service with an SLA, Reached by Identity | static-validated |
| Google Cloud | A Vertex AI Search Data Store in Your Region, Encrypted with Your Key, with Its Engine | static-validated |
| Huawei Cloud | A Cloud Search Service Cluster with Security Mode On, Which Is What Makes a Password Mean Anything | static-validated |
| IBM Cloud | A Private Elasticsearch Deployment with Three Members, Because Two Cannot Elect | static-validated |
| Oracle Cloud | An OpenSearch Cluster that Authenticates and Has a Quorum | static-validated |
| Tencent Cloud | An Elasticsearch Cluster with Authentication On, Three Public Switches Closed and COS Backups | static-validated |
How to choose
Decide first whether you are sizing a cluster or paying per query: a provisioned domain is cheaper at steady load and needs a node count, while serverless bills by compute units and scales to zero traffic but not to zero cost. Then compare where the endpoint answers from (a VPC, or the internet behind IAM) and whether retention is a policy or a job you write.
When not to use
Search is a derived copy of your data, not the system of record. Losing the index costs a reindex, not the data; treat retention and snapshots as a cost control rather than a backup.