While I really liked the neat API and it's easy to use, the index performance and ram/disk size is really disappointing in a use case I oversee.
Typesense in my opinion is much better on performance (ymmv but in my case it's ~15% disk/ram, 5% of index time comparing to Meili). However, it's a bit behind on easy to use: the auto schema api has some limitation, also lack of official arm64 build (maybe it's a c++ vs rust toolchain thing, it takes quite some time to compile by oneself, with many c++ libs used from different vendors like google/baidu/etc, and some lib even requires SSE)
It's fun to see both come with comparison charts [1] [2], I dunno who started this but competition is good for the industry (except I'm mildly concerned GitHub stars are used in PR like this), and I hope both get better over time.
This is Kishore, co-founder of Typesense. We will be publishing an ARM64 build from the next release. And I hear you about the auto schema: working on that as well.
Congrats! Love seeing an alternative to Elastic/Lucene, esp. in Rust! Just wondering, for those using Elasticsearch for log storage/querying, could Meilisearch potentially fulfill this use-case also? And are you aiming to be a more "performant" and efficient alternative to Elasticsearch?
Thanks kevinsf90 :D
Meilisearch is really focus towards `end-user search` or `customer-facing search` so, there is few chances we go in that direction.
But we are really aiming to be the easier, simpler and thus more performant alternative to Elasticsearch. Since Meilisearch won't cover every uses cases that Elasticsearch covers, it is much more performant and intuitive on this subset.
If you want to build complex queries, with terabytes of data & aggregation queries you should definitely use Elasticsearch. If you want to build the best search experience for your end-user that will be the most relevant and answers in a few milliseconds, then Meilisearch is the obvious choice here :)
Meilisearch is rather the free and open alternative to Algolia, as it focuses on solving zero-config and easy to deploy autocomplete search, as opposed to analytics at scale.
Typesense in my opinion is much better on performance (ymmv but in my case it's ~15% disk/ram, 5% of index time comparing to Meili). However, it's a bit behind on easy to use: the auto schema api has some limitation, also lack of official arm64 build (maybe it's a c++ vs rust toolchain thing, it takes quite some time to compile by oneself, with many c++ libs used from different vendors like google/baidu/etc, and some lib even requires SSE)
It's fun to see both come with comparison charts [1] [2], I dunno who started this but competition is good for the industry (except I'm mildly concerned GitHub stars are used in PR like this), and I hope both get better over time.
[1] https://docs.meilisearch.com/learn/what_is_meilisearch/compa...
[2] https://typesense.org/docs/overview/comparison-with-alternat...