Erick Erickson has been in the programming business for over (ahem) 37 years. He started working with Lucene in 2008, replacing several different legacy systems with Lucene. In 2010 he began working for Lucid Imagination (a.k.a. Lucidworks) helping clients implement and manage Solr installations, starting with Solr 1.4. Since 2012, Erick has been an independent consultant working with clients large and small to solve a wide range of real-world, search-related problems. He's seen the evolution of Solr from a stand-alone search engine where 50 million documents required significant care and feeding to modern Solr, where billions of documents are routine. Erick has observed that we're still struggling with hardware constraints, it's just the scale of the problem has vastly expanded; we tackle problems that we ignored not so long ago because we knew we didn't have the hardware. Along the way, Erick became a Lucene/Solr committer and PMC member, actively engaged in the Lucene/Solr ecosystem.
Erick Erickson is speaking at the following session/s
Modifying Solr indexes Without Re-Ingesting
With some exceptions, our advice is usually "whenever you change the Solr schema, you should reindex all your data". Reindexing all the data for a large corpus can range from easy to tedious to impossible. There are some situations in which the index could be modified to conform to the new schema without reingesting all of the data from the system-of-record. This talk will outline the mechanism by which this may be done, one collection at a time, even with collections sharing a common configset in SolrCloud.
An understanding of the challenges of updating an index without re-indexing, as well as a high-level view of how to implement the required processes/programs.
Solr developers and anyone who must maintain Lucene/Solr indexes and has been faced with "you must re-index from scratch" for whom that process is difficult/impossible.