Group By Counts


(Gustavo Arjones-2) #1

Hi,

I'm evaluating replace MySQL+Fulltext search for ES.

Currently I fetch information from twitter, using twitter stream api
and save it to mysql, of course, this solution isnt scalable ...

ES seems to be solution for my needs, there is one point I couln't
figure out: it is how to get count() on a date range grouped by date -
I use this kind of information to show some trends.

My query is something like:

Select created_at, count()
From summary
where create_at between '2010-01-01' and '2010-01-31'
and
match(text) against ('java OR ruby' in boolean match)
Group by create_at;

What should be the best way to replicate this behavior?

Thanks a lot!
Gustavo Arjones


(Shay Banon) #2

Check out the facets support elasticsearch (the histogram facet), in
upcoming 0.9.1 there will also be a range facet that supports dates.

-shay.banon

On Thu, Aug 19, 2010 at 4:04 AM, Gustavo Arjones
gustavo.arjones@gmail.comwrote:

Hi,

I'm evaluating replace MySQL+Fulltext search for ES.

Currently I fetch information from twitter, using twitter stream api
and save it to mysql, of course, this solution isnt scalable ...

ES seems to be solution for my needs, there is one point I couln't
figure out: it is how to get count() on a date range grouped by date -
I use this kind of information to show some trends.

My query is something like:

Select created_at, count()
From summary
where create_at between '2010-01-01' and '2010-01-31'
and
match(text) against ('java OR ruby' in boolean match)
Group by create_at;

What should be the best way to replicate this behavior?

Thanks a lot!
Gustavo Arjones


(system) #3