I need a database that supports fast sequential reads in lexicographical order, like LevelDB. This is usually achieved by using a structure like a B-tree. But then I would like it to be distributed, ideally being able to rebalance itself amongst existing and new nodes as data grows. Does such a thing exist?