Migrate from Pinecone to Zilliz Cloud

This topic describes how Zilliz Cloud handles data type mapping, field conversion, namespace processing, and collection naming rules when migrating from Pinecone.

Prerequisites

Before starting your Pinecone to Zilliz Cloud migration, ensure you meet these requirements:

Pinecone requirements

Requirement	Details
Index type	Supports migrating from Pinecone Serverless indexes only
API access	Pinecone API key with access permissions
Data availability	Source indexes from Pinecone must contain data. Empty indexes cannot be migrated.
Vector dimension	Dimension must be > 1. Single-dimension vectors will cause migration failure

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

Understanding how Pinecone data types map to Zilliz Cloud is crucial for planning your migration:

Pinecone Field Type	Zilliz Cloud Field Type	Notes
Primary key	VARCHAR (primary key)	Automatically mapped. Enable Auto ID to generate new IDs (original values will be discarded).
Dense vector	FLOAT_VECTOR	Dimensions preserved exactly, no modifications needed
Sparse vector	SPARSE_FLOAT_VECTOR	Only mapped if non-empty in sample data.
Metadata	Dynamic fields	Mapped as dynamic schema by default; can be converted to fixed fields. Refer to Dynamic Field for more details.
Namespace	Partition key / partition	Recommended for performance optimization. Refer to Namespace processing for more details.

Metadata field conversion

📘Notes

Zilliz Cloud samples 100 rows to detect metadata schema. You can manually add additional fields if needed.

Pinecone metadata is initially mapped to Zilliz Cloud's dynamic schema for maximum flexibility. You can optionally convert metadata fields to fixed fields to gain:

• Enforced data types for stronger validation

• Optimized indexing for better query performance

• Structured schema for consistent data management

When converting metadata to fixed fields:

Pinecone Metadata Type	Zilliz Fixed Field Type	Notes
String	VARCHAR	Maximum 65,535 bytes supported
Number (int/float)	DOUBLE	All numeric types become DOUBLE
Boolean	BOOL	Direct mapping
List of strings	ARRAY<VARCHAR>	Nested arrays supported

For metadata fields converted to fixed fields, you can configure additional attributes:

• Nullable: Decide whether a field can accept null values. This feature is enabled by default. For details, refer to Nullable attribute.

• Default Value: Set fallback values when data is missing. For details, refer to Default values.

Pinecone-specific handling rules

Namespace processing

Pinecone namespaces can be migrated using two strategies:

Strategy	Implementation	Performance Impact	Use Case
Namespace as Partition Key (Recommended)	Namespaces become values in a partition key field	Automatic optimization for search performance	Most scenarios with multiple namespaces
Namespace as Partition	Each namespace becomes a separate partition	Manual partition management required	Simple scenarios with few, stable namespaces

📘Notes

Pinecone's default namespace handling:

• As Partition: Becomes _default partition in Zilliz Cloud

• As Partition Key: Becomes empty string "" value

For more information on partition and partition key concepts, refer to Manage Partitions and Use Partition Key.

Collection naming rules

Pinecone index names are automatically processed for Zilliz Cloud compatibility:

Pinecone Index Name	Zilliz Cloud Collection Name	Rule Applied
my-vector-index	my_vector_index	Hyphens (-) converted to underscores (_) to comply with Zilliz Cloud collection naming conventions
product_search	product_search	No change needed

Naming conflicts: If a collection with the same name already exists in the target database, you must:

• Delete the existing collection, or

• Choose a different target database, or

• Rename the target collection during migration configuration