Which statement is true regarding Azure Cosmos DB's indexing feature?

The statement that indexing can be applied even when the structure varies from record to record is true regarding Azure Cosmos DB's indexing feature. Azure Cosmos DB is designed as a schema-agnostic database, allowing it to handle a wide variety of data types and structures, which makes it particularly suited for applications that require flexibility in data representation.

This flexibility means that different records within the same collection can have vastly different attributes or structures. Given this capability, Azure Cosmos DB can automatically create indexes on the different attributes of varying records without requiring a predefined schema. This allows developers to easily query diverse datasets without needing to conform all data to a single structure in advance, enhancing both performance and user experience when dealing with unpredictable or evolving data types.

In contrast, the other statements either imply limitations or restrictions that do not apply to Azure Cosmos DB. For example, the fact that non-relational databases can have indexing is generally supported in systems like Azure Cosmos DB, refuting the notion that indexing is exclusive to relational databases or that data must be of the same format to be indexed. Hence, the ability to index varied structures dynamically is one of the key features that differentiate Azure Cosmos DB and contributes to its versatility in handling unstructured or semi-structured data.