Github user simonellistonball commented on the issue:
https://github.com/apache/metron/pull/945
Are we losing anything by moving the scheme from Range to Trie types?,
repeating my comment on
https://github.com/apache/metron/pull/922:
Given that our use case is heavily dependant on sorting, I wonder why not
the Trie based indices for numeric fields. I may be completely wrong on their
advantages but would love to hear the logic behind the choice of Point indices.
If there is a good reason, maybe we should consider retaining those for 6.6
in addition to the 5.5 clusters. Either way it would be be good to understand
the basis for the type decision.
---
