Denis, I absolutely agree with you that a SortedSet doesn't need to be partitioned. The main idea I wanted to put attention on is that now it's not possible to find a row with a particular number with log complexity. We need to scan node-local index until we reach some offset, it seems to be linear. For example Redis uses a combination of a hash table and skip list to achieve that. I'm not aware how complicated it is to implement it in Ignite as we use trees. Just wanted to draw some attention to it.
-- Sent from: http://apache-ignite-developers.2346864.n4.nabble.com/
