Why can't they be? I've seen indexing documents on the order of 6k/second. Is the problem with getting the data out of the DB (i.e. how long does the SQL take to execute) or on the Solr indexing side? If the former, you could consider writing a SolrJ program that accessed the database and sent the docs to Solr. If you can write the program such that you could run N copies (each updating only a portion of the data) you could run them in parallel.
But I suspect that the issue is probably your SQL, and before looking at Solr solutions I'd try to verify that. Best, Erick On Fri, Nov 23, 2012 at 11:03 AM, zbindigonzales <sandrozbin...@gmail.com>wrote: > Hi there. > > In our company we use Apache Solr 4 to index data from the database via the > databaseimport handler. > The data we are indexing is a denormalzied table (Patient, Visit, Study, > Image). > > One requirement is to be near realtime. For that we use softcomits every > second. > The index size is about 3M documents. > > The problem now is that if a useres modify a entry on a the patient table > there may be 1000 rows affected (due to denormalization) wich can't be > indexed in near realtime. > > Did anyone had a similar problem and got a workarround to speed up delta > imports ? > > Best Regards > > > > > > > > > > > -- > View this message in context: > http://lucene.472066.n3.nabble.com/Solr-Near-Realtime-with-denormalized-Data-tp4022072.html > Sent from the Solr - User mailing list archive at Nabble.com. >
