In our current implementation, we store each query as a file to DFS. We should reduce the overhead.
In our current implementation, we store each query as a file to DFS. We should reduce the overhead.