Is there some roadmap for 2i features? We are currently using search for indexing as we need to perform compound queries. After some testing I found that if we could use i2 we probably see 4-5x improvements in insert throughput and and 1.5-2x in querying times, depending on the query. We can work around the lack of an intersection operation (AND), by implementing the filtering in a MR phase. My testing shows that is sufficiently performant in many cases. But we are in a bind by the lack of a union operation (OR). Executing multiple individual exact matches introduces too much latency given that we may have hundreds of exact matches in a single query.
A more general solution to the lack of a union operation would be to allow multiple "inputs" statements to a MR job, although the the job may have to perform some deduplication or just accept dups, depending on how the inputs are structured. If would also be rather handy if the key filter language could be applied to matching secondary indexes. Elias
_______________________________________________ riak-users mailing list riak-users@lists.basho.com http://lists.basho.com/mailman/listinfo/riak-users_lists.basho.com
