Consider Typesense as an alternative to OpenSearch for search functionality

[ilmimris]

Hi OpenMetadata team,

I wanted to open a discussion about potentially supporting Typesense as an alternative search backend alongside or instead of OpenSearch.

Context:
Currently, OpenMetadata uses OpenSearch for search and discovery capabilities. While OpenSearch is powerful, I think Typesense could offer some compelling advantages for certain deployment scenarios.

Potential benefits of Typesense:

• Significantly lighter resource footprint and faster deployment
• Simpler configuration and maintenance compared to OpenSearch
• Built specifically for instant search with typo tolerance out of the box
• Better performance for typical metadata search use cases
• Lower operational complexity for smaller to medium deployments
• Native support for semantic search and vector operations
• More cost-effective for organizations that don't need full ELK stack capabilities

Questions:

• Has the team considered Typesense as an option?
• Are there technical or architectural reasons why OpenSearch was chosen over lighter alternatives like Typesense?
• Would the community be interested in a contribution that adds Typesense support as an alternative backend?
• What would be the team's concerns or requirements for such an integration?
• Are there specific OpenSearch features that OpenMetadata heavily relies on that might not have equivalents in Typesense?

I'd be happy to discuss implementation approaches if there's interest from the maintainers and community.

Thanks for considering this suggestion!