Give it a try. If there is anything wrong, please point it out. There are also relatively few related articles. Some are my own guesses
First go to the impala query page of cdh. View query details
Then look at the following parameters
Duration: Literally the duration of the query
HDFS scanner average read throughput: hdfs io speed
Number of backeds: Query the number of instances used
Peak memory usage of each node: multiplied by the number of instances is the memory used by this query
CPU usage should be thread:CPU time divided by duration
Number of HDFS bytes read: literally