On 4/5/06, jason rutherglen <[EMAIL PROTECTED]> wrote: > Is anyone looking at partitioning Solr indices over several machines and then > querying using a multisearcher? Or creating a multisearcher like class for > Solr? Would this idea be considered with the design of Solr or a bad idea? >
It's definitely come up, and would be a good idea for certain applications. Partitioning by request type or document types would be ideal since Solr can already work unchanged in that capacity (the partitioning smarts go in a proxy or something). A MultiSearcher type of approach can also make sense, but is much more difficult. IMO, it's probably a bad idea unless your index is simply too large to service queries in a reasonable amount of time. Things become much harder when distributed (for example, how would you take the intersection of two sets of documents in a distributed manner)? Any type of distributed approach would be limited to a subset of functionality. -Yonik
