Re: Best way to handle multi-billion row read-only table?

2010/2/10 Greg Stark <gsstark(at)mit(dot)edu>:
> On Wed, Feb 10, 2010 at 2:32 PM, Asher <asher(at)piceur(dot)co(dot)uk> wrote:
>> The data will initially be accessed via a simple GUI which will allow
>> browsing over a subset of the data (subsampled down to 1 sample/minute/hour,
>> etc.
>
> It sounds like you could use a tool like rrd that keeps various levels
> of aggregation and intelligently chooses the right level for the given
> query. I think there are such tools though I'm not sure there are any
> free ones.

Use as much memory as possible to fit indexes as well as portions of
the table space itself in RAM.
Of course, poor indexing can kill any effort.

--
Vincenzo Romano
NotOrAnd Information Technologies
NON QVIETIS MARIBVS NAVTA PERITVS