Hello Rahul
Blind guess, classic facet may pick `method=enum`, where `json.facet` might
be more straightforward. Can you check debugQuery=true response and tweak
`method` for `json.facet` ?


On Thu, Jan 29, 2026 at 10:23 PM Rahul Goswami <[email protected]>
wrote:

> Hello all,
> I am seeing a large performance drawdown on an index (8 shards) with around
> 1 billion docs with json.facet. The fields are docValues true, and only 11
> docs qualify for the query. But the facet part itself takes 3 to 4 times
> the time taken by the query part (aka without facet).
>
> Questions:
> - Does json.facet in solr work differently than regular facets
> (facet=true&facet.field=myfield, etc)?
>
> - Is one more performant than the other for a large index or under certain
> constraints?
>
> I should mention that the faceting is being done on couple of high
> cardinality fields. But still, it *feels *like json.facet is considering a
> much larger dataset for evaluation even though only 11 docs qualify.
>
> Thanks,
> Rahul
>


-- 
Sincerely yours
Mikhail Khludnev

Reply via email to