lucene-java-user mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Karl Wettin <karl.wet...@gmail.com>
Subject Re: Caching in lucene
Date Tue, 18 Sep 2007 07:23:53 GMT

18 sep 2007 kl. 08.53 skrev Shailendra Mudgal:

> Ok let me explain you. By warming up the reader i used to  
> understand is that
> it memorizes the index terms. Therefore subsequent queries will be  
> answered
> by using these terms. Is this correct ?
>

Are you asking if all terms are flyweighted/interned by Lucene? The  
string values they hold are, but I'm not sure about the actual  
instances of Term.

Warming is merely the process of avoiding latency by using a reader a  
bit prior to allowing consumers to access it. It does not include any  
extra caching layers. Solr does <http://wiki.apache.org/solr/ 
SolrCaching>.


-- 
karl

>
>
> On 9/18/07, Karl Wettin <karl.wettin@gmail.com> wrote:
>>
>>
>> 18 sep 2007 kl. 08.33 skrev Shailendra Mudgal:
>>
>>> Whether it caches frequently used terms ?
>>
>> I don't understand you question?
>>
>> --
>> karl
>>
>>
>>>
>>>
>>> On 9/18/07, Karl Wettin <karl.wettin@gmail.com> wrote:
>>>>
>>>>
>>>> 18 sep 2007 kl. 07.12 skrev Shailendra Mudgal:
>>>>
>>>>>
>>>>> What my goal is to understand the caching strategy.  How well this
>>>>> work for repetitive queries. Is there any room available to  
>>>>> improve
>>>>> this.
>>>>
>>>> It is usually a loss of resources to cache results in a busy system
>>>> with gaussianity distributed queries. There is a paragraph on the
>>>> wiki <http://wiki.apache.org/lucene-java/LuceneCaveats> that  
>>>> explains
>>>> this: Lucene is amazingly fast at searching. Rather than caching  
>>>> hits
>>>> and paging through them, merely re-executing the query is often  
>>>> fast
>>>> enough.
>>>>
>>>> I have conducted experiements with active results cache at index
>>>> update time, soft/weak references and some other things I can't  
>>>> think
>>>> of right now. Most of the time it ended up with the garbage  
>>>> collector
>>>> going crazy with zillions of tiny soft/weak referenced instances,
>>>> clock ticks spent updating results people don't that much, et c.  
>>>> You
>>>> really need a lot of repetative queries in order for such a  
>>>> cache to
>>>> do any good.
>>>>
>>>> If you are retrieving a lot of stored fields it might make sense to
>>>> cache documents that are frequently accessed.
>>>>
>>>>
>>>> --
>>>> karl
>>>>
>>>>
>>>>
>>>> ------------------------------------------------------------------- 
>>>> --
>>>> To unsubscribe, e-mail: java-user-unsubscribe@lucene.apache.org
>>>> For additional commands, e-mail: java-user-help@lucene.apache.org
>>>>
>>>>
>>
>>
>> ---------------------------------------------------------------------
>> To unsubscribe, e-mail: java-user-unsubscribe@lucene.apache.org
>> For additional commands, e-mail: java-user-help@lucene.apache.org
>>
>>


---------------------------------------------------------------------
To unsubscribe, e-mail: java-user-unsubscribe@lucene.apache.org
For additional commands, e-mail: java-user-help@lucene.apache.org


Mime
View raw message