Hi,
We are currently investigating some performance issues we have related to
Solr's graph query parser (i.e. {!graph ...}).
In our index with test data we have 11,111 root documents. At the next level of
each root document we have 40 documents, and at the next and final level we
have 5 documents. So when we build a graph for one root document it contains a
total of 241 documents.
Building the graphs for all 11,111 root documents down to the leaf nodes takes
about 26.5 seconds, which is slow.
The total amount of documents for all of these graphs is 2,677,751 which isn't
a lot.
Is there a way to make it faster? Maybe some caching feature that we've missed?
We've tried increasing the available heap space to 8 GiB but it didn't make a
difference.
Best regards,
Tobias
