Migrate from OpenSearch to Zilliz Cloud
This topic describes how Zilliz Cloud handles data type mapping, collection naming rules, and considerations when migrating from OpenSearch.
Prerequisites
Before starting your OpenSearch to Zilliz Cloud migration, ensure you meet these requirements:
OpenSearch requirements
| Requirement | Details |
|---|---|
| Network access | Source OpenSearch cluster must be accessible from the public internet |
| Authentication | Valid cluster endpoint, username, and password with necessary permissions |
| Vector field requirement | Each source index must contain at least one k-NN vector field |
| Data availability | Source indexes must contain data. Empty indexes cannot be migrated. |
Zilliz Cloud requirements
| Requirement | Details |
|---|---|
| User role | Organization Owner or Project Admin |
| Cluster capacity | Sufficient storage and compute resources (use the CU calculator to estimate CU size) |
| Network access | Add Zilliz Cloud IPs to allowlists if using network restrictions |
Data type mapping
The following table summarizes how field types in OpenSearch are mapped to Zilliz Cloud field types, along with details on any customization options.
| OpenSearch Field Type | Zilliz Cloud Field Type | Description |
|---|---|---|
| Primary key | Primary key | OpenSearch's primary key (_id) is automatically mapped as the primary key in Zilliz Cloud. When migrating data, you can enable Auto ID. However, if you do so, the original primary key values from your source table will be discarded. |
| k-NN vector | FLOAT_VECTOR | The float vector type from OpenSearch is mapped to FLOAT_VECTOR on Zilliz Cloud. Byte/Binary vectors from OpenSearch are not supported for migration.Vector dimensions remain unchanged. |
| Alias | Not supported | Alias fields are not supported. |
| Binary | VARCHAR | Binary data is stored as a string on Zilliz Cloud. |
| Numeric | ||
byte | INT8 | Directly mapped. |
double | DOUBLE | Directly mapped. |
float | FLOAT | Directly mapped. |
half_float | FLOAT | Mapped to FLOAT. |
integer | INT32 | Directly mapped. |
long | INT64 | Directly mapped. |
short | INT16 | Directly mapped. |
unsigned_long | Not supported | Not supported on Zilliz Cloud. |
scaled_float | Not supported | Not supported on Zilliz Cloud. |
| Boolean | BOOL | Stores true or false. |
| Date | VARCHAR | Stored as a string. Ensure correct format conversion. |
| IP address | VARCHAR | Stored as a string. |
| Range | JSON | Stored in JSON format. |
| Object | ||
object | JSON | Stored in JSON format. |
nested | JSON | Stored in JSON format. |
flat_object | JSON | Stored in JSON format. |
join | VARCHAR | Stored as a string. |
| String | ||
keyword | VARCHAR | Stored as a string. |
text | VARCHAR | Mapped to VARCHAR . |
match_only_text | VARCHAR | Stored as a string. |
token_count | INT32 | Stored as INT32. |
wildcard | Not supported | Not supported on Zilliz Cloud. |
| Autocomplete | VARCHAR | Stored as a string. |
| Geographic | VARCHAR | Stored as a string. |
| Rank | VARCHAR | Stored as a string. |
| Percolator | VARCHAR | Stored as a string. |
| Derived | Not supported | Derived fields are not supported on Zilliz Cloud. |
| Star-tree | Not supported | Star-tree fields are not supported on Zilliz Cloud. |
| Arrays | Not supported | Arrays are not supported for migration. |
| Multifields | Not supported | Multifields are not supported for migration. |
OpenSearch-specific handling rules
Collection naming rules
OpenSearch index names are transferred to Zilliz Cloud with the following considerations:
| Scenario | Impact | Solution |
|---|---|---|
| Default naming | Collection names match source index names exactly | Names are preserved as-is from OpenSearch |
| Special characters | Index names with hyphens (-) or dots (.) will cause errors and prevent job submission | Manually rename indexes to use underscores or other valid characters |
| Naming conflicts | Cannot submit job if a collection with the same name already exists | Delete existing collection, choose a different database, or rename during migration configuration |
Migration considerations
The following features are not supported for OpenSearch migration:
| Limitation | Impact | Alternative |
|---|---|---|
| Dynamic to fixed field conversion | Cannot convert existing dynamic fields to fixed types | Fields maintain their original dynamic nature |
| Add more fields | Cannot add new fields during migration | Only existing Elasticsearch fields are migrated |
| Sparse vectors | Not supported in current release | Consider dense vector alternatives or contact support for roadmap |