I would recommend that you search the list for some
great discussions on NDFS. Doug has a nice writeup of
his vision of using a map reduce job to push the
indexes to your query servers so they're updates as
the webdb is and managed that way.

NDFS just wasn't designed for the I/O of a query. You
want to have queries either cached in memory or on
fast disk drives or else the latency of reading over a
network would slow things down terribly especially if
you have concurrent queries.

-byron

--- Gal Nitzan <[EMAIL PROTECTED]> wrote:

> So just use the ndfs command to download the
> relevant files from NDFS 
> and put them on the search server and from there to
> follow the sample on 
> your documentation project?
> 
> Thanks for all the help.
> 
> P.S. Do you have a clear view for the solution to
> the "slowness in 
> search over NDFS"? if so I would be interested in
> giving a big hand on 
> that. Since it is a crucial part for my company...
> 
> Regards,
> 
> Gal
> 
> Stefan Groschupf wrote:
> > There will be a solution soon, if I found some
> more time, until this 
> > for smaller installation you need a shell script
> that download the 
> > index and segment to the box that runs the search
> server.
> > You also can move the index from ndfs to local
> instead of copy it.
> > check: "bin/nutch ndfs" for documentation.
> >
> > Am 28.12.2005 um 14:34 schrieb Gal Nitzan:
> >
> >> Whoa, that was fast...
> >>
> >> So all in all you would need two sets of the same
> data?
> >>
> >> Did I understand there is an effort to improve
> the "poor performance" 
> >> issue?
> >>
> >> And if we are at it, would you care to explain
> how to download the 
> >> index to local and what happens if the data is
> growing over the 
> >> boundaries of one machine? do you just add HD to
> the machine?
> >>
> >> Thanks,
> >>
> >> Gal
> >>
> >> Stefan Groschupf wrote:
> >>> Download index to a local file system.
> >>>
> >>> Am 28.12.2005 um 14:25 schrieb Gal Nitzan:
> >>>
> >>>> Hi,
> >>>>
> >>>> If using search over NDFS is too slow than what
> is the alternative 
> >>>> when all your data is in NDFS?
> >>>>
> >>>> Thanks, Gal
> >>>>
> >>>>
> >>>
> >>>
> >>> .
> >>>
> >>
> >>
> >>
> >
> >
> > .
> >
> 
> 
> 

Reply via email to