How does atlas search compare against Elastic search features?

I am basically evaluating Atlas search to migrate our workload from ES.

Do you have any feature by feature comparison against ES. Or how most common things used in ES can be setup in Atlas search.
pipeline_database.json (2.5 KB)
template_database.json (4.3 KB)

Currently I am looking to map ingest pipelines and templates from ES to Atlas search but cannot find any good place where I can look into regarding this topics.

there are some features of Elasticsearch that we do not have, and others that our users do not need. Index templates is not something we have released yet.

For example, alias in your index definition is not something you need anymore because we have a transparent feature called no downtime indexing.If you change your index, we will swap it out for the old one and it will be accessible at the same name.

1 Like

This feature request for partial indexing might be one that you will be interested in. We will do it and index templates in sprints in the near future, and you will receive updates about our progress if you vote here: Support Partial Indexing – MongoDB Feedback Engine

1 Like

Hi @Marcus, thanks … do you have any docs that I can refer to regarding how atlas search works under the hood?
We have dynamic indexes created based on dates and the index are sharded. So currently that cannot be achieved with atlas search I guess?
Also how does the perf of atlas search compare to ES, given that indexes can be sharded in ES?
Since Atlas is using lucene under the hood, does atlas search provide any runtime knobs for tuning for eg for JVM?

  1. What are the restrictions and performance characteristics of deep pagination in Atlas Search?
  2. would it be possible to transform document before indexing like in ES pipeline?
  3. Disable norms norms | Elasticsearch Guide [8.6] | Elastic and doc_values doc_values | Elasticsearch Guide [8.6] | Elastic – not sure but I think they are mapped to Lucene settings?


  1. There is a private preview of the improved performance for deep pagination. It will be released soon, but the behavior of the data store changes quite a bit when users enable it so we want to ensure customers know how to use it appropriately. Although, I always recommend for any search engine that you should work on improving your relevancy if you need to do deep pagination. For application search, we would be interested in the use case and desired behavior.

  2. You can use MongoDB Atlas Triggers.

  3. The option for norms is in the string index definition documentation as a parameter.

1 Like