Re:Ask HN: What's the best way to handle full-text search wi...
by rlivsey on Tuesday January 04, 2011 @08:51PM
(#2069460)
We're using ElasticSearch [1] and finding it really nice to work with.
It's basically Lucene with a nice JSON/RESTful interface and distributed out of the box.
Unlike most other Lucene based systems that I've played with it allows you to get started without having to define your schema upfront, so you can just chuck data at it and worry about field types later.
It's under constant development and the community is really helpful.
Re:Ask HN: What's the best way to handle full-text search wi...
by rlivsey on Wednesday January 05, 2011 @07:03AM
(#2070719)
We've not put it into production yet (it's coming in the new version) but are putting it through its paces in staging and it's handling things well so are confident it's ready for our usage.
There are people on the mailing list which are using it for some fairly hefty things.
Usual case of YMMV and as with anything it's worth taking a bit of time to try out the various options and see how they stack up for your usage. A good thing about ES is that it's so quick and easy to get running that trying it out and doing some benchmarking is pretty trivial.
Re:Ask HN: What's the best way to handle full-text search wi...
by pierrefar on Wednesday January 05, 2011 @04:42AM
(#2070506)
It's really easy: Sphinx has a data import interface called xmlpipe2 which takes an XML stream/file and feeds it to Sphinx for indexing. So I wrote a small PHP command line script that gets all the documents I want indexed from MongoDB and formats them as XML. I gave a talk about it last year:
The other aspect of coure is configuring Sphinx. That really depends on what you're doing and want from it. The forums are helpful and the documentation and wiki are superb. Really they are - read them!
It's basically Lucene with a nice JSON/RESTful interface and distributed out of the box.
Unlike most other Lucene based systems that I've played with it allows you to get started without having to define your schema upfront, so you can just chuck data at it and worry about field types later.
It's under constant development and the community is really helpful.
[1] - http://www.elasticsearch.com/
There are people on the mailing list which are using it for some fairly hefty things.
Usual case of YMMV and as with anything it's worth taking a bit of time to try out the various options and see how they stack up for your usage. A good thing about ES is that it's so quick and easy to get running that trying it out and doing some benchmarking is pretty trivial.
http://skillsmatter.com/podcast/cloud-grid/mongodb-full-text...
The other aspect of coure is configuring Sphinx. That really depends on what you're doing and want from it. The forums are helpful and the documentation and wiki are superb. Really they are - read them!