How does elastic compare to mongodb for vector search

Elastic Stack Elasticsearch
1

we are evaluating elastic and mongodb for our vector search usecase. What are some advantages of elastic over mongodb?

2

I honestly don't know what Mongo is providing in this space and anyway I'm biased :slight_smile:

I'd just say that Elasticsearch IS a search engine and is known for that and to be extremely powerful. I'm seeing Mongo more as a datastore with some search features.

Some of the key features I love with Elasticsearch:

  • Quantization from float to int8, int4 and bit.
  • Same _search API as you are using for lexical search, including standard support for pagination, aggregation, filters, sort
  • Because of it, it's easy to mix token based search and vector search

Just naming a few... There's plenty of goodness in this space.
Have a look at Elasticsearch Labs to get even more information on what's cooking...

1 Like
3

I found with Mongodb you’re less able to flexibly combine clauses of various types due to the way they use multiple engines under the covers.
I don’t know how this might apply to vector support but I know this is true of plain old text search - see search - Can MongoDB query (A AND B) OR (C AND D) with any mix of text and structured clauses? - Stack Overflow

1 Like