Selective indexing at namespace level

For pod-based indexes, Pinecone indexes all metadata by default. Use selective metadata indexing to prevent this

Request 1: would be highlight this somewhere in starter notebooks, as many aren’t aware of this issue until they face query performance issues. Its like you index all columns of database table unless you explicitly specify. Coming from world of SQL & NoSQL databases, people will ideally expect index to be created only on demand.

Request 2: would be to have the capability of Selective metadata indexing at namespace level, rather than whole of index. Bottleneck here is, querying can be done at namespace level only, but metadata indexing should be maintained at index level. That frustrating.

Hi @infra and welcome to the Pinecone community forums!

Thank you for your feedback - we really appreciate it.

Out of curiosity, when you’re referring to starter Notebooks, do you mean the Notebooks we have here: GitHub - pinecone-io/examples: Jupyter Notebooks to help you get hands-on with Pinecone vector databases?

Or are you referring to something else?

I’ll log your feedback so that it’s seen by the proper team.


I’ve logged your feedback and feature request internally.


Just in case you weren’t aware, the Serverless architecture does away with the need for selective metadata, and any metadata fields can be filtered on or returned in the queries. Are you using the pod-based indexes for something you don’t think is a good fit for Serverless?