I want to run 3 different types of queries on Hive LLAP and want to assign resources for each query type. Something like creating different queues in Yarn and setting min and max capacities. Is it possible to do that? If possible, I would want to enable preemption as well.
I am trying to understand more about concurrent users and sessions in LLAP. I understand on Hive:
number of concurrent users = number of default queues * number of sessions per queue
What is the difference between number of concurrent sessions and number of concurrent users? Also, for Hive LLAP, number of default queues is always 1 (LLAP runs only on llap queue), so does the above equation still hold? For a cluster we need to set up, it has at least 20 concurrent users. (Total users is around 50). We have 8 data nodes and gave 90% of the cluster for LLAP. How should I set up resources so everyone is able to run their queries, every has some resources for themselves?