Known limitations
This page describes known limitations and feature restrictions in Pinecone.
General
- Upserts
-
Pinecone is eventually consistent, so there can be a slight delay before upserted records are available to query.
After upserting records, use the
describe_index_stats
operation to check if the current vector count matches the number of records you expect, although this method may not work for pod-based indexes with multiple replicas. -
Only indexes using the dotproduct distance metric support querying sparse-dense vectors.
Upserting, updating, and fetching sparse-dense vectors in indexes with a different distance metric will succeed, but querying will return an error.
-
Indexes created before February 22, 2023 do not support sparse vectors.
-
- Metadata
- Null metadata values are not supported. Instead of setting a key to hold a null value, remove the key from the metadata payload.
Serverless indexes
Serverless indexes are in general availability on AWS and in public preview on GCP and Azure. Check the serverless limits and restrictions.
Serverless indexes do not support the following features:
-
Migrating pod-based indexes to serverless indexes
- Serverless migration is limited to pod-based indexes with less than 25 million records and 20,000 namespaces.
- The pod-based indexes must be hosted on AWS or GCP. It is not currently possible to migrate pod-based indexes on Azure.
- The migration process creates serverless indexes on AWS only, regardless of the cloud where the pod-based indexes are hosted. This means that pod-based indexes on GCP will be migrated to serverless indexes on AWS.
-
- Instead, you can delete records by ID prefix.
-
- Because high-cardinality metadata in serverless indexes does not cause high memory utilization, this operation is not relevant.
-
- Because serverless indexes scale automatically based on usage, this operation is not relevant
-
- This feature is available on AWS only.
Pod-based indexes
- Pod storage capacity
- Each p1 pod has enough capacity for 1M vectors with 768 dimensions.
- Each s1 pod has enough capacity for 5M vectors with 768 dimensions.
- Metadata
- Metadata with high cardinality, such as a unique value for every vector in a large index, uses more memory than expected and can cause the pods to become full.
- Collections
- You cannot query or write to a collection after its creation. For this reason, a collection only incurs storage costs.
- You can only perform operations on collections in the current Pinecone project.
- Sparse-dense vectors
- Only
s1
andp1
pod-based indexes using the dotproduct distance metric support sparse-dense vectors.
- Only
Was this page helpful?