Re: Partitioning the index

[Jim Adams <ja...@gmail.com>]

Are there any particular suggestions on memory size for a machine?  I have a
box that has only 1 million records on it - yet I'm finding that date
searches are already unacceptable (30 seconds) slow.  Other searches seem
okay though.

Thanks!

On Thu, Dec 18, 2008 at 2:02 PM, Yonik Seeley <ys...@gmail.com> wrote:

> It's more related to how much memory you have on your boxes, how
> resource intensive your queries are, how many fields you are trying to
> facet on, what acceptable response times are, etc.
>
> Anyway... a single box is normally good for between 5M and 50M docs,
> but can fall out of that range (both up and down) depending on the
> specifics.
>
> -Yonik
>
> On Wed, Dec 17, 2008 at 9:34 PM, s d <s....@gmail.com> wrote:
> > Hi,Is there a recommended index size (on disk, number of documents) for
> when
> > to start partitioning it to ensure good response time?
> > Thanks,
> > S
> >
>

Re: Partitioning the index

[Jim Adams <ja...@gmail.com>]

that is what I thought.  Thanks.

On Tue, Jan 6, 2009 at 10:18 PM, Shalin Shekhar Mangar <
shalinmangar@gmail.com> wrote:

> You'll need to re-index.
>
> On Wed, Jan 7, 2009 at 9:49 AM, Jim Adams <ja...@gmail.com> wrote:
>
> > It's a range query.  I don't have any faceted data.
> >
> > Can I limit the precision of the existing field, or must I re-index?
> >
> > Thanks.
> >
> > On Tue, Jan 6, 2009 at 8:41 PM, Yonik Seeley <ys...@gmail.com> wrote:
> >
> > > On Tue, Jan 6, 2009 at 10:06 PM, Jim Adams <ja...@gmail.com>
> wrote:
> > > > Are there any particular suggestions on memory size for a machine?  I
> > > have a
> > > > box that has only 1 million records on it - yet I'm finding that date
> > > > searches are already unacceptable (30 seconds) slow.  Other searches
> > seem
> > > > okay though.
> > >
> > > I assume this is a date  range query (or date faceting)?
> > > Range queries with many unique terms in the range is a known
> > > limitation, and we should hopefully fix this in 1.4.
> > > In the meantime, limiting the precision of dates could help a great
> deal.
> > >
> > > -Yonik
> > >
> >
>
>
>
> --
> Regards,
> Shalin Shekhar Mangar.
>

Re: Partitioning the index

[Shalin Shekhar Mangar <sh...@gmail.com>]

You'll need to re-index.

On Wed, Jan 7, 2009 at 9:49 AM, Jim Adams <ja...@gmail.com> wrote:

> It's a range query.  I don't have any faceted data.
>
> Can I limit the precision of the existing field, or must I re-index?
>
> Thanks.
>
> On Tue, Jan 6, 2009 at 8:41 PM, Yonik Seeley <ys...@gmail.com> wrote:
>
> > On Tue, Jan 6, 2009 at 10:06 PM, Jim Adams <ja...@gmail.com> wrote:
> > > Are there any particular suggestions on memory size for a machine?  I
> > have a
> > > box that has only 1 million records on it - yet I'm finding that date
> > > searches are already unacceptable (30 seconds) slow.  Other searches
> seem
> > > okay though.
> >
> > I assume this is a date  range query (or date faceting)?
> > Range queries with many unique terms in the range is a known
> > limitation, and we should hopefully fix this in 1.4.
> > In the meantime, limiting the precision of dates could help a great deal.
> >
> > -Yonik
> >
>



-- 
Regards,
Shalin Shekhar Mangar.

Re: Partitioning the index

[Jim Adams <ja...@gmail.com>]

It's a range query.  I don't have any faceted data.

Can I limit the precision of the existing field, or must I re-index?

Thanks.

On Tue, Jan 6, 2009 at 8:41 PM, Yonik Seeley <ys...@gmail.com> wrote:

> On Tue, Jan 6, 2009 at 10:06 PM, Jim Adams <ja...@gmail.com> wrote:
> > Are there any particular suggestions on memory size for a machine?  I
> have a
> > box that has only 1 million records on it - yet I'm finding that date
> > searches are already unacceptable (30 seconds) slow.  Other searches seem
> > okay though.
>
> I assume this is a date  range query (or date faceting)?
> Range queries with many unique terms in the range is a known
> limitation, and we should hopefully fix this in 1.4.
> In the meantime, limiting the precision of dates could help a great deal.
>
> -Yonik
>

Re: Partitioning the index

[Yonik Seeley <ys...@gmail.com>]

On Tue, Jan 6, 2009 at 10:06 PM, Jim Adams <ja...@gmail.com> wrote:
> Are there any particular suggestions on memory size for a machine?  I have a
> box that has only 1 million records on it - yet I'm finding that date
> searches are already unacceptable (30 seconds) slow.  Other searches seem
> okay though.

I assume this is a date  range query (or date faceting)?
Range queries with many unique terms in the range is a known
limitation, and we should hopefully fix this in 1.4.
In the meantime, limiting the precision of dates could help a great deal.

-Yonik