This should be OK. There's no real issue with Solr docs having up to 200 fields, and theres no real limitation on what portion of those fields each doc has. In other words, only having 20 out of 200 possible fields in a doc isn't a problem. There's no overhead for "unused" fields.
Depending upon the number of *unique* values in these fields, you may bet some extra memory consumption, but whether that matters depends a lot on the particulars. By a huge margin, most of the time is spent in searching itself, not parsing the field names so long ones really don't matter at all I think. Best Erick On Tue, Jan 31, 2012 at 7:18 AM, tech20nn <vijaypawnar...@gmail.com> wrote: > We are planning to import data from various tables of ERP DB into a single > Solr/Lucene index. > Since these tables have overlapping columns we are planing to name the > corresponding document field as <table_name>_<column_name>. I have > following questions on this. > > 1) Does having long field name (<table_name>_<column_name>) affect > performance ? > 2) We will end up with close to 200 fields per document in schema > definition. At time of document storage only 20 field per will be indexed > and stored for each stored document. Is there a limitation here, are we > creating performance bottleneck by designing schema this way ? > > Thanks > Vijay > > > > -- > View this message in context: > http://lucene.472066.n3.nabble.com/Document-with-longer-field-names-and-many-fields-tp3703077p3703077.html > Sent from the Solr - User mailing list archive at Nabble.com.
