@cdklabs/generative-ai-cdk-constructs
@cdklabs/generative-ai-cdk-constructs / s3vectors / VectorIndexProps
Properties for creating a VectorIndex resource
readonlyoptionaldataType:FLOAT_32
The data type of the vectors to be inserted into the vector index
- FLOAT_32
readonlydimension:number
A dimension is the number of values in a vector. A larger dimension needs more storage.
All vectors added to the index must have exactly this number of values. Must be an integer between 1 and 4096.
readonlyoptionaldistanceMetric:VectorIndexDistanceMetric
The distance metric to be used for similarity search
- COSINE
readonlyoptionalencryption:VectorIndexEncryption
The kind of server-side encryption to apply to this index.
If you choose KMS, you can specify a KMS key via encryptionKey. If
encryption key is not specified, a key will automatically be created.
KMS if encryptionKey is specified, or S3_MANAGED otherwise.
readonlyoptionalencryptionKey:IKey
External KMS key to use for index encryption.
The encryption property must be either not specified or set to KMS.
An error will be emitted if encryption is set to S3_MANAGED.
encryption is set to KMS and this property is undefined,
a new KMS key will be created and associated with this index.
readonlyoptionalnonFilterableMetadataKeys:string[]
Non-filterable metadata keys allow you to enrich vectors with additional context during storage and retrieval. Unlike default metadata keys, these keys can’t be used as query filters. Non-filterable metadata keys can be retrieved but can’t be searched, queried, or filtered. You can access non-filterable metadata keys of your vectors after finding the vectors.
- All metadata attached to vectors is filterable and can be used as filters in a similarity query
readonlyvectorBucket:IVectorBucket
The vector bucket to use for the vector index
readonlyoptionalvectorIndexName:string
The name of the vector index
- Assigned by CloudFormation (recommended).