lucene-solr-user mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Susheel Kumar <susheel2...@gmail.com>
Subject Re: Sharding vs single index vs separate collection
Date Fri, 09 Jun 2017 02:11:30 GMT
correction: shared => sharded

On Thu, Jun 8, 2017 at 10:10 PM, Susheel Kumar <susheel2777@gmail.com>
wrote:

> You mentioned most of the searches will use document routing based on year
> as route key, correct? and then you mentioning  huge amount of searches
> again without routing.  Can you give some no# how many will utilise routing
> vs not routing?
>
> In general, we should try to serve all the queries with one
> index/collection which can be shared if needed or replicated to serve huge
> amount of queries.  Having a separate index should be avoided unless you
> have very good reasons.
>
> Thnx
>
>
>
> On Thu, Jun 8, 2017 at 5:45 PM, Johannes Knaus <Knaus@mpdl.mpg.de> wrote:
>
>> Hi,
>> I have a solr cloud setup, with document routing (implicit routing with
>> router field). As the index is about documents with a publication date, I
>> routed according the publication year, as in my case, most of the search
>> queries will have a year specified.
>>
>>
>> Now, what would be the best strategy -as regards performance (i.e. a huge
>> amount of queries to be processed)- for search queries without any year
>> specified?
>>
>> 1 - Is it enough to define that these queries should go over all routes
>> (i.e. route=year1, year2, ..., yearN)?
>>
>> 2 - Would it be better to add a separate node with a separate index that
>> is not routed (but maybe sharded/splitted)? If so, how should I deal with
>> such a separate index? Is it possible to add it to my existing Solr cloud?
>> Would it go into a separate collection?
>>
>> Thanks for your advice.
>>
>> Johannes
>
>
>

Mime
  • Unnamed multipart/alternative (inline, None, 0 bytes)
View raw message