I thought it was odd that there was mention of competitors but then when a feature matrix ranking them many were suddenly omitted. Then I rechecked the URL and realized this was a marketing post by milvus.
I have no skin in or say in the vector db space but do support a very large (1.4B+ vector) Vespa deployment and see 40ms p99 query times. YMMV but it seems like people sleep on Vespa for whatever reason. Again, not an endorsement but this is what we’re seeing in prod at a fortune100 co.
With a product like this that costs so much and is so deeply technical, no one is buying on emotion, so to me, omitting data is tantamount to admitting their product is inferior in some way.
I thought it was odd that there was mention of competitors but then when a feature matrix ranking them many were suddenly omitted. Then I rechecked the URL and realized this was a marketing post by milvus.
I have no skin in or say in the vector db space but do support a very large (1.4B+ vector) Vespa deployment and see 40ms p99 query times. YMMV but it seems like people sleep on Vespa for whatever reason. Again, not an endorsement but this is what we’re seeing in prod at a fortune100 co.
With a product like this that costs so much and is so deeply technical, no one is buying on emotion, so to me, omitting data is tantamount to admitting their product is inferior in some way.
vector search solutions mentioned in this post:
Milvus
Qdrant
Weviate
Open Search
Pgvector
Redis
Cassandra
Solr
Vespa
Pinecone
Vertex AI