OpenSearch:Application schema in OpenSearch High-performance Search Edition

Full-text search

Use these types for fields that need tokenization and multiple analysis methods.

Exact match

Use these types for fields queried by exact value — filtering, sorting, or joining on a precise string.

Numeric

If your data source fields are FLOAT or DOUBLE type, change them to DECIMAL to avoid data precision issues.

Temporal

Fields of the INT type can also be mapped to a DATETIME or TIMESTAMP field in a data source. The values are automatically converted to milliseconds. Use the range function to filter results by time range.

Geographic

Array types

Array field types (INT_ARRAY, FLOAT_ARRAY, DOUBLE_ARRAY, LITERAL_ARRAY) follow the same general pattern:

• In a data source, map the field to a VARCHAR or STRING column. Use the MultiValueSpliter data processing plug-in to convert it.

• When uploading via the OpenSearch API or SDKs, send the field as an array, not a string.

// Correct: array
String[] literal_array = {"Alibaba Cloud", "OpenSearch"};

For DOUBLE_ARRAY fields in ApsaraDB RDS and PolarDB data sources, only VARCHAR is supported (not STRING). For MaxCompute, both VARCHAR and STRING are supported.

Reserved field names

Reserved names: service_id, ops_app_name, inter_timestamp, index_name, pk, ops_version, ha_reserved_timestamp, summary

Field value limits

Each field type enforces value range limits. If a value falls outside the allowed range, an overflow error occurs or the value is truncated. For the full list, see the "Limits on fields" section of Limits.

Supported field types by data source

Field type mapping reference