Optimizing SparQL Queries in AWS Neptune to Avoid Memory Issues

Hi guys! Good evening, someone knows how can I make a more efficient SparQL query to sample some IDs than this:

SELECT DISTINCT ?sourceId WHERE {
     ?sourceId :hasX ?object.
}
ORDER BY RAND()
LIMIT 100

It's returning out of memory errors inconsistenly, seems that the smaller the limit more likely to get out of memory(???), running in AWS Neptune.