Is this an admission that you can use your existing databases instead of buying into the specialized db trend?
My understanding is that the embedding and comparison metric are what is important, and that you can use plugins like pg_vector to extend the applications of your database (not sure how this is for other db ecosystems). Curious to hear what people more in the guts of these things have to say?
This article got corrected a day after it was posted probably because there was significant backlash from Pinecone. It turns out Pinecone only stores user info about their free tier users on Google Cloud. I can't imagine they have more than few thousand customers on Google Cloud. So it is a very small dataset that could run easily on any database service and it has nothing to do with vector embeddings. Pinecone probably stores the same metadata for their customers on AWS using RDS or Aurora. And they surely aren't boasting about it.
Not sure what's going on with Spanner. They are getting called out on one misleading marketing post after another recently.
2 comments
[ 3.1 ms ] story [ 10.7 ms ] threadMy understanding is that the embedding and comparison metric are what is important, and that you can use plugins like pg_vector to extend the applications of your database (not sure how this is for other db ecosystems). Curious to hear what people more in the guts of these things have to say?
Not sure what's going on with Spanner. They are getting called out on one misleading marketing post after another recently.