CODEWITHSUNDEEP
solrlucenenosql
Johnny000
Johnny000

Reputation: 2104

Solr join best approach

I have a scenario where I have a productdatabase in solr and a branddatabase in MySql. In the solr productdatabase I have a field named brandid where I refer to the Mysql primary key from the branddatabase. Now I would like to join the branddatabase for each solr searchquery and groups the result seperatly from the product results. I thought about a second solr database where I save the branddata and then join it on every query, but I would like to have each brand only one time and not merged together with the product results in the same resultset. A facette-style result for the brands is my goal. Anyone has a pointer how I could achieve this kind of results in my xml/json?

The resultset how I would like to have it in pseudo solr code:

<results>
    <products>
        <product>
            ...
        </product>
        <product>
            ...
        </product>
        <product>
            ...
        </product>
        <product>
            ...
        </product>
    </products>
    <brands>
        <brand>
            ...
        </brand>
        <brand>
            ...
        </brand>
    </brands>
</results>

Upvotes: 0

Views: 189

Answers (2)

Greg
Greg

Reputation: 617

Can you use a higher-level language?

I currently do something similar, but I use Java as the glue. The Java application takes in requests, goes against solr using solrj, retrieves all the results, including the facets, I take that response and query against mysql to get more information, I merge all the data in the java layer and then construct the xml/json response.

solrj

other higher-level languages are offered:

Ruby,PHP,Java,Scala,Python,.Net,Perl,Javascript

Upvotes: 1

Ion Cojocaru
Ion Cojocaru

Reputation: 2583

If you only need to serve additional fields from brand database and you do not need to search/filter on them then you could apply a simple faceting on brandid and populate the presentation fields in a post processing step from DB directly/memory cache/key value store... and use facet.mincount=1 to eliminate the brands without any products in the current query.

Upvotes: 3

Related Questions