Since the change from Oracle Text to Apache Solr/Lucene 1.4 in DC-X, time has passed and Solr 1.4 is now considered “old” software. (To be fair, it is a very stable version, running successfully on our customer’s DC-X systems.) Solr 1.4 has been succeeded by the 3.x branch.

The current release 3.6 contains a lot of improvements. Not only major bugfixes (affecting some DC-X systems), but also enhancements in memory allocation and minor performance improvements drive us to Solr 3.6. One highlighting is an improved index segment merge strategy, which makes time consuming optimizations obsolete. I.e., because merges of segments are optimized, the index won’t be fragmented so fast. This is especially worthwhile for systems with lots of data flowing through them, such as DC-X with news agency data and RSS feeds.

A new functionality is the JOIN feature which is similar to a database join. This can help to get relevant information depending on the particular queries. Such as a list of themes for a specific publication. One main reason to move to Solr 3.0 will be the fact that Solr itself only supports the index version from one before. And DC is planning to move very soon to the upcoming Solr 4.0 for DC-X, where we expect major performance improvements. So it can beware from reindexing which stresses system and support personnel…

Thomas Wülfken
About Thomas Wülfken

Thomas is leading the Support team. He started working for DC in 1996 and specialized in Oracle (OCP), MySQL and PostgreSQL.

Leave a Reply