Quantiles require accessing the entire list of results, or at least, sorting by the interesting values, checking the total hits, then accessing the results list at the desired interval. So, with 3000 hits, get deciles by getting the first row, then the 301st row, the 601st row, etc.
This might be slow and require a lot of memory. Solr is optimized for showing the top few results. Anything that accesses all results or even requests rows far down the list can be very slow. If your main use requires accessing all results, Solr may not be the right choice. A relational database is designed for efficient operations over the entire set of results. wunder On Mar 29, 2012, at 11:50 PM, Kashif Khan wrote: > Hi all, > > I am doing R&D about SOLR using any quantiles function for a set. I need a > quick-start road-map for modifying that quantiles function in my SOLR > plugin. I am thinking that it might be using any third party tools or > library for it. > > -- > Kashif Khan > > -- > View this message in context: > http://lucene.472066.n3.nabble.com/Quantiles-in-SOLR-tp3870084p3870084.html > Sent from the Solr - User mailing list archive at Nabble.com.
