Collections in FAST do not exist in Solr. A FAST collection can be implemented in Solr using facets or shards. The collection abstraction in FAST is actually more shard-like in semantics: it is a separate top-level set of content. This has strong ramifications for relevance: if collections have the same relevance "statistical footprint", they can go in the same shard. If they have different relevance characteristics they should go in different shards.
Example: if book collections and movie title collections share one shard, relevance calculations are completely bogus. They should go into 2 separate shards and with different search tuning. I did one conversion during Solr 1.2. I would up mass-editing all of the XML data files into Solr'd XML input format. I cannot recommend this technique. (Note: since FAST charges by query, doing a deep walk and uploading to Solr was not financially feasible.) In general, expect to do a test conversion and then redesign your schema and search strategies for your "real" conversion. Solr has a lot of subtleties. On Tue, Nov 24, 2009 at 8:11 AM, Tommy Molto <tommymo...@gmail.com> wrote: > This is really a great source of migration. I guess i will have good > questions after trying. But what i know that will be a little harder will be > the use of collections (facets in Solr) and hierarquical navigators. > > On Tue, Nov 24, 2009 at 1:05 PM, Shashi Kant <sk...@sloan.mit.edu> wrote: > >> Here is a link that might be helpful: >> >> http://sesat.no/moving-from-fast-to-solr-review.html >> >> The site is choc-a-bloc with great information on their migration >> experience. >> >> >> On Tue, Nov 24, 2009 at 8:55 AM, Tommy Molto <tommymo...@gmail.com> wrote: >> >> > Hi, >> > >> > I'm new at Solr and i need to make a "test pilot" of a migration from >> Fast >> > ESP to Apache Solr, anyone had this experience before? >> > >> > >> > Att, >> > >> > -- Lance Norskog goks...@gmail.com
