bq. I am trying to see how sharding can be employed to improve the query performance by adding the route to a shard based on a field in schema.xml.
This is actually straightforward, just create the collections with implicit routing. See the collections API for CREATE and the "Document Routing" section of the reference guide. That said, for an index that size I suspect you have other problems and splitting up to shards isn't a long-term solution. I regularly see 200G indexes in the wild (admittedly some fairly beefy machines) that return sub-second response times. FWIW, Erick On Mon, Jul 2, 2018 at 7:43 PM, Sushant Vengurlekar <svengurle...@curvolabs.com> wrote: > We have two collections which are 21G and constantly growing. The index on > one of them is also 12G. I am trying to see how sharding can be employed to > improve the query performance by adding the route to a shard based on a > field in schema.xml. So I am trying to figure out how to split the > collections into shards based on this one field and then query them further > by routing the query to a particular shard based on this field. > > Thank you > > On Mon, Jul 2, 2018 at 7:36 PM, Erick Erickson <erickerick...@gmail.com> > wrote: > >> This seems like an "XY problem". _Why_ do you want to do this? >> Has your collection outgrown one shard and you feel you have to >> split it? Sharding should only be used when you can't host your >> entire collection on a single replica and still get adequate performance. >> >> When you do reach that point, the usual process is to just >> decide how many shards you need and let Solr do the rest >> of the work. Why do you think you need to specify how docs >> are routed based on some field? >> >> Best, >> Erick >> >> On Mon, Jul 2, 2018 at 6:06 PM, Sushant Vengurlekar >> <svengurle...@curvolabs.com> wrote: >> > I want to split a collection based on one field. How do I do it and then >> > query based off that. >> > >> > Ex: collection1. Field to split off col1 >> > >> > Thank you >>
