lucenenet-user mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Trevor Watson <twat...@datassimilate.com>
Subject Re: Optimization times
Date Mon, 24 Jan 2011 20:08:56 GMT
Thanks to all for the speedy and useful replies!

On 01/24/2011 2:20 PM, Kevin Miller wrote:
> I battled with How To Optimize Lucene a bit. Make sure you understand why
> you are calling Optimize. You only need to optimize to improve search
> performance (by limiting the number of index files which get traversed when
> collecting hits). The index writer will "optimize" automatically based on
> your MergeFactor settings. IMHO your in-flight optimizations are likely
> unnecessary. I also found that it was best to use the Optimize(int) overload
> which to select the minimum number of index files (sorry I forgot the
> correct term) which sped up optimization quite a lot as getting a large
> index into a single index file can be quite time consuming.
>
> I found this post very educational regarding this subject:
> http://tim.oreilly.com/pub/a/onjava/2003/03/05/lucene.html?page=1
>
> On Mon, Jan 24, 2011 at 12:58 PM, Trevor Watson
> <twatson@datassimilate.com>wrote:
>
>> Just a quick question regarding using the Optimize function in Lucene.NET.
>>
>> Is it more time efficient to call Optimize occasionally while adding
>> documents to an index, or is it better to call it at the end of adding
>> documents only?
>>
>> The index we are creating has a possible 2-3 million records added at a
>> time and we currently optimize every 100,000.
>>
>> Thanks in advance.
>>
>> Trevor Watson
>>


Mime
View raw message