Github user kkhatua commented on the issue: https://github.com/apache/drill/pull/755 Thanks, @vrozov. I'll make use of a separate lock for read-only purpose in case of `#1`. For `#2`, I need to construct a size-limited ordered set from a list of unordered elements. In this case, the elements (i.e. profiles) need to be ordered by file-name, which is a 1:1 mapping function of the start time epoch for the query. So, I need to be able to add to such a datastructure in `O(log(n))` time, remove in `O(1)` and iterate through it in sequence. So, my puts are the most expensive operation.
---