https://issues.apache.org/jira/browse/SOLR-8492 shows an example of the AnalyticsQuery where the merge is being handled by the Streaming API. I actually think this is nicer then then using MergeStrategy. The Streaming API gives you full control over the merge from the shards.
Joel Bernstein http://joelsolr.blogspot.com/ On Wed, Mar 16, 2016 at 12:49 PM, sudsport s <sudssf2...@gmail.com> wrote: > Hi , > > I am planning to write custom aggregator in solr which will use some > probabilistic data structures per shard to accumate results and then after > shard merging results will be sent to user as integer. > > I explored 2 options to do this > > 1. Solr analytics API > https://cwiki.apache.org/confluence/display/solr/AnalyticsQuery+API > > I can implement merge policy and post filter to perform aggregation , I > have example working using this , but I am not sure if it is ok to pass > objects which > 1 MB in shard response? > does solr use javabin serialization to optimize data gathering from shards? > then leader shard will collect these 1 MB probabilistic data structures & > produce count which will be included in response. > > > 2. JSON Facet API http://yonik.com/json-facet-api/ > > After looking at > > https://github.com/apache/lucene-solr/tree/master/solr/core/src/java/org/apache/solr/search/facet > > FacetProcessor.java seems very similar to Solr analytics API. > seems like merging happens similar way where response will include objects > like hll and merge them > > > > > > one key difference is Solr analytics API is based on postFilter and JSON > facet API is based on ValueSource but I dont understand impact of using one > or the other. > > > Can someone help me out? >
