lucene-java-user mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Adrien Grand <jpou...@gmail.com>
Subject Re: MultiReader question
Date Thu, 18 Aug 2016 11:20:17 GMT
I don't think there would be significant disadvantages of using
MultiReader, but depending what the data looks like, there might not be
benefits either. If data is homogeneous per entity but not across entities
then the MultiReader approach might have potential for making things more
efficient, but otherwise it would perform just like a monolithic index that
has more segments (since merging works per IndexWriter).

Le jeu. 18 août 2016 à 11:28, Cristian Lorenzetto <
cristian.lorenzetto@gmail.com> a écrit :

> Ipothesis:
> i want split universe set to index in different subtopics/entities/subsets
> Considering for semplicity entity X is indexed in Index_X folder.
>
> Considering now i have a query when is searched
> X1,X2,X3 ,
> for semplicity N is the number of sub readers.
>
> I could use MultiReader lucene api for making it.
>
>
> questions:
>
> 1) what about MultiReader performance /memory  in relation to a monolitic
> index?  memory used is N times what used by a monolitic index?
> 2) avantages /disadvantages MultiReader against SingleReader?
>

Mime
  • Unnamed multipart/alternative (inline, None, 0 bytes)
View raw message