Re: SOLR 1.4 and 1.3 diff and other

2008-12-17 Thread Sammy Yu
Hi Yonik,
   Thanks for the quick response.  Do you know the release schedule
when 1.4 would be released or if it is possible to backport the NIO
implementation into 1.3?  If you could give me a pointer that would be
great.  It seems like a huge performance gain that would be of value
to a lot of people.

Thanks,
Sammy

On Wed, Dec 17, 2008 at 5:36 PM, Yonik Seeley  wrote:
> On Wed, Dec 17, 2008 at 7:52 PM, Sammy Yu  wrote:
>> I read somewhere that there are contention issues with the current
>> cache implementation of LRUCache in 1.3 in that it is synchronous,
>> could this be the reason why the filter query are slow?
>
> Probably not.  The change is much more likely due to using a
> non-blocking NIO implementation in Lucene.
>
> -Yonik
>


Re: SOLR 1.4 and 1.3 diff and other

2008-12-17 Thread Yonik Seeley
On Wed, Dec 17, 2008 at 7:52 PM, Sammy Yu  wrote:
> I read somewhere that there are contention issues with the current
> cache implementation of LRUCache in 1.3 in that it is synchronous,
> could this be the reason why the filter query are slow?

Probably not.  The change is much more likely due to using a
non-blocking NIO implementation in Lucene.

-Yonik


SOLR 1.4 and 1.3 diff and other

2008-12-17 Thread Sammy Yu
Hi,
   I'm making a simple query that uses the standard query handler to
make constructed query such as
title:iphone OR text:firmware.  Next, I am use the filterquery to
limit the amount of items to data from within the last year via
fq=+dateCreated:[NOW-1YEAR/MONTH TO NOW/MONTH] which is significantly
faster than putting it as part of the query since the filter cache is
populated and almost all my queries have need this filter.  So my
filter cache is always 1 and it always just contains this 1 entry.

That said this result doesn't seem to scale very well in SOLR 1.3,
when I hit around 10 simultaneous threads, the response time jumps
over 200 msec.  Without the filter query parameter and just the query
straight up, I can handle up to 26 threads.   I was kind of blown away
when I ran the same test on SOLR 1.4 nightly build with the fq
parameter, it handled 26 ms in less than 140 ms.

I read somewhere that there are contention issues with the current
cache implementation of LRUCache in 1.3 in that it is synchronous,
could this be the reason why the filter query are slow?  If so are
there ways to use the newer caching strategies from 1.4 and backport
it into 1.3 as a cache plugin?  I would like the 1.4 performance
gains, but we are hesitant to roll out something that hasn't been
deemed release quality.

Thanks,
Sammy