I posted back on SOLR-7090 <https://issues.apache.org/jira/browse/SOLR-7090>. I'm probably misunderstanding it, but what's there seemed pretty rough to me.
On Tue, Sep 29, 2015 at 7:32 PM, Scott Blum <[email protected]> wrote: > On Fri, Sep 25, 2015 at 6:49 PM, Erick Erickson <[email protected]> > wrote: > >> yeah, the streaming stuff is pretty bleeding-edge but pretty cool. >> > > I looked at this for a bit, but I wasn't clear on the performance > implications of streaming. We use pagination heavily in our solr > interactions, and the doc seems to suggest this isn't supported with > streaming? That it's really designed with the export case in mind for data > analysis rather than real time queries. Am I reading that wrong? > > >> Your understanding is accurate, the pathological case is the reason >> it's not been implemented in core Solr. I suppose you could do exactly >> what you outlined, just with two queries. >> >> for SOLR-4095, why would this affect sharding for your main collection? >> The groups collection is just a separate collection, I don't see why you >> think it would affect sharding of the main collection. That just means I >> don't understand your problem probably... >> > > Ah, no it's just that we have a multi-tenant environment where we have a > collection per tenant (on the order of many hundreds). Putting the group > info into a side collection would mean doubling the number of collections > (and the effort into managing them). And just in general it would > complicate our general indexing and searching code. Not impossible, but > we're rather avoid it. > >
