Unfortunately, we have about 8 different fields that could serve as 
aggregation key, and a lot of potential combinations between these fields.
Thus, pre-computing all these combinations doesn't seem to be a viable 
solution.

On Friday, January 31, 2014 7:52:40 AM UTC-8, Binh Ly wrote:
>
> Maxime, your bottleneck is likely in the script part. It has to 
> dynamically compute that per doc just like in sql. However, if you can 
> precompute that at index time (for example, introduce a field that contains 
> the value of date-hour-id, you should be able to improve that aggregation 
> time significantly. I did a quick test in 1.0 RC1 with an index of about 
> 100K docs, and if I precompute that term field (and eliminate the script 
> part), it is at least 10x faster than the script version. YMMV.
>

-- 
You received this message because you are subscribed to the Google Groups 
"elasticsearch" group.
To unsubscribe from this group and stop receiving emails from it, send an email 
to [email protected].
To view this discussion on the web visit 
https://groups.google.com/d/msgid/elasticsearch/134a71d9-7683-4804-9ae9-449d40580b35%40googlegroups.com.
For more options, visit https://groups.google.com/groups/opt_out.

Reply via email to