This document tells you (nearly) all you need to know, everyone should read it:

http://developer.marklogic.com/inside-marklogic

Ellis.

On 23 Sep 2013, at 15:35, "Ganyushkin, Andrey - Moscow" <[email protected]> 
wrote:

>> each document that matches will be retrieved and scanned through to find the 
>> matches
> 
> It means that indexes refer to documents, not to concrete tags in the 
> documents. And each time MarkLogic scans documents (if we drop cache). It's 
> correct?
> 
> For example. If I have two expressions
> 
>    First - fn:collection()//ns1:itemX
> 
>    Second - 
> fn:collection()/ns1:root/ns1:lavel1/ns1:level2/ns1:level3/ns1:itemX
> 
> The second expression will be more faster if each document contains a lot of 
> different tags? (because MarkLogic needs time for scan all tags in the 
> document)
> 
> Thanks,
> Andrey
> 
> -----Original Message-----
> From: [email protected] 
> [mailto:[email protected]] On Behalf Of Damon Feldman
> Sent: Monday, September 23, 2013 5:16 PM
> To: MarkLogic Developer Discussion
> Subject: Re: [MarkLogic Dev General] full search or not
> 
> Andrey,
> 
> Yes, it will find that element wherever it exists in XML documents in the 
> database. All XML structure is always indexed in MarkLogic at the document 
> level, so the indexes can tell MarkLogic Server which documents contain 
> ns1:item at all. To find every element within those documents, each document 
> that matches will be retrieved and scanned through to find the matches. 
> 
> Adding indexes in MarkLogic is somewhat different from adding indexes in 
> other systems because there are these two kinds of indexes: the Universal 
> Index which is always on, and Range Indexes (a form of lexicon) which is 
> explicitly defined per element. Range indexes are usually for fast sorting, 
> facets and other operations beyond normal searching, so you can search for 
> XML structures, words and the like without any special index settings and it 
> will be quite fast.
> 
> Note that searching for every value in a database can return huge amounts of 
> data, however, so you may need to use another technique to limit the results.
> 
> Yours,
> Damon
> 
> --
> Damon Feldman
> MarkLogic Corporation
> www.marklogic.com
> 
> 
> -----Original Message-----
> From: [email protected] 
> [mailto:[email protected]] On Behalf Of Ganyushkin, 
> Andrey - Moscow
> Sent: Monday, September 23, 2013 3:54 AM
> To: MarkLogic Developer Discussion
> Subject: [MarkLogic Dev General] full search or not
> 
> Hi All,
> 
> If I use the following expression
> 
> //ns1:item
> 
> It's mean that ML scan all items (all tags and sub tags) in the my DB or not?
> and if I put ns1:item in DB index, could this speed up the query?
> 
> Please explain it to me.
> 
> Thanks,
> Andrey
> 
> 
> _______________________________________________
> General mailing list
> [email protected]
> http://developer.marklogic.com/mailman/listinfo/general
> _______________________________________________
> General mailing list
> [email protected]
> http://developer.marklogic.com/mailman/listinfo/general
> 
> 
> _______________________________________________
> General mailing list
> [email protected]
> http://developer.marklogic.com/mailman/listinfo/general
_______________________________________________
General mailing list
[email protected]
http://developer.marklogic.com/mailman/listinfo/general

Reply via email to