I am running a single query multiple times on LLAP,I get to see a different execution time for each run. Since LLAP uses in-memory cache the execution time should go down. But I get the following results for my query.for each run(23s,14s,49s,43s,36s,13s.........)
The query I run is an Insert select query, Which spawns 43 mappers and 30 reducers tasks. I only see the difference in the mapper time and the reducer time remains constant around 4secs.
Hi. Can you please provide details of your environment. Are you running on cloud or dedicated hardware? What are the specs of memory an I/O for your nodes? Are other applications being run on your cluster at the same time? Shared by other users? Thank you.