On 5/17/2013 2:47 AM, Carlos Bonilla wrote: > To calculate some stats we are using a field "B" with 50.000.0000 different > values as facet pivot in a schema that contains 200.000.000 documents. We > only need to calculate how many different "B" values have more than 1 > document but it takes ages.... Is there any other better way/configuration > to do this? > > Configuration: > Solr 4.2.1 > JVM Java 7 > Max Java Heap size : 12Gb > 8 GB RAM > Dual Core
You probably don't have enough RAM. With 200 million documents, I would imagine that your index is considerably larger than 4GB in size. With the 16GB of RAM that you mentioned in your other message, this configuration leaves 4GB of RAM for caching after Java manages to allocate the entire 12GB heap - which it will do very quickly with a large index. See the following: http://wiki.apache.org/solr/SolrPerformanceProblems I don't know the size of your index. If it is 100GB, then ideally you would want to have at least 112GB of RAM, but you could probably make it work in 64GB. Thanks, Shawn
