S3 vector will fundamentally change the landscape of vector DB.
from our previous evaluation pg_vector is already good enough, but it has limitation of scalability and hosting issues. S3 is built-in with scalability beyond a single node.
Cheaper vector storage/querying due to AWS’s economics of scale coupled with their elasticity/reliability? Guess I’m migrating to S3 Vectors.
crazy how we’ve already memoryholed the vector db mania from 2023
hot take: fomo-driven speculative venture capital bubbles are not an efficient way to allocate our citizens’ pension money
It will be a good fit for object data you already have in S3, but it won’t fit every use case.
Good retrieval needs a hybrid approach to get the best results.
