search

couchbase / cbft

Couchbase Full Text server
Other
137 stars 28 forks source link

still under developing? or achieved? #10

Open maxn0d3x opened 5 years ago

maxn0d3x commented 5 years ago

Looks like there is no 1.0 version.. and the last release was in 2015 ! As I know, cbft is the core full text search engine of couchbase, right? I am curious why this project is getting more inactive even though this engine is the core feature of the Couchbase product ! Could you explain about couchbase's full text search engine status for now and about cbft ?

keshavmr commented 5 years ago

CBFT is very much under continuous development, along with the bleve project. CBFT is part of Couchbase server and released as the integrated product. The last release was 6.0.1, and stay tuned for a beta release coming up later in 2Q2019. Here's a sample of the feature we're working for search: https://blog.couchbase.com/n1ql-and-search-how-to-leverage-fts-index-in-n1ql-query/

maxn0d3x commented 5 years ago

@keshavmr Oh my goodness! 👍 I am now ready to replace Super heavy freaking jvm-based ElasticStack. I don't like Java/JVM anymore..It is too heavy and memory consuming. I even hate it. I am thinking of that couchbase/cbft with bleve can be a good alternative of Elasticsearch/mongodb.

You think Couchbase fulltext search performance is almost same like Elasticsearch in some specific case ? not all because Elastic search is full-featured (that is why it is heavy). My main concern is high performance with tons of millions documents in a specific case. Such as searching products in ecommerce db or text articles in cms.

It would be great to contribute my technology for couchbase stack. !

keshavmr commented 5 years ago

@NiNJAD3vel0per You've correctly identified the sweet spot. Most applications have "database" workload and some "search" workload. Instead of deploying to different products, you can use a single product (Couchbase) to achieve your outcome.

CBFT (FTS) should meet the performance (with Couchbase 6.0 or higher) for the typical use case you mention. We'd be interested to hear any of your feedback.