Thank you Shawn!
So if I run the two following requests, it will only store once 7.5Mo,
right?
- select?q=*:*&fq=bPublic:true&rows=10
- select?q=field:my_search&fq=bPublic:true&rows=10
kr,
Bast
On 04/05/2016 16:22, Shawn Heisey wrote:
On 5/3/2016 11:58 PM, Bastien Latard - MDPI AG wrote:
Thank you for your email.
You said "have big caches or request big pages (e.g. 100k docs)"...
Does a fq cache all the potential results, or only the ones the query
returns?
e.g.: select?q=*:*&fq=bPublic:true&rows=10
=> with this query, if I have 60 millions of public documents, would
it cache 10 or 60 millions of IDs?
...and does it cache it the filter cache (from fq) in the OS cache or
in java heap?
The result of a filter query is a bitset. If the core contains 60
million documents, each bitset is 7.5 million bytes in length. It is
not a list of IDs -- it's a large array of bits representing every
document in the Lucene index, including deleted documents (the Max Doc
value from the core overview). There are two values for each bit - 0 or
1, depending on whether each document matches the filter or not.
Thanks,
Shawn
Kind regards,
Bastien Latard
Web engineer
--
MDPI AG
Postfach, CH-4005 Basel, Switzerland
Office: Klybeckstrasse 64, CH-4057
Tel. +41 61 683 77 35
Fax: +41 61 302 89 18
E-mail:
lat...@mdpi.com
http://www.mdpi.com/
