Navin, Your specification of 40 concurrent users and data size are only a bit less than half the story. Without the rest of the story, nobody will be able to give you even general guidance beyond a useless estimate that it will take between roughly 1 and 40 drillbits with with a gob of memory.
To do better than such non-specific "guidance", you need to add some additional answers. For example, What is the query mix? How long do these queries run without any question of concurrency? Could that query speed be enhanced with better partitioning? How are you storing your data? What promises are you making to these concurrent users? On Mon, Apr 13, 2020 at 7:21 AM Navin Bhawsar <navin.bhaw...@gmail.com> wrote: > Hi Team , > > We are planning to use drill to query hdfs cluster with about a terabyte > data in parquet file format .There will be approx. 40 concurrent users > using this environment . > > Can you please advise on below two points considering above workload for > optimum performance: > 1. Number of Drill bits > 2. Memory Config per drill bit > > Also is there any sizing guide for Apache Drill to refer. > > > Thanks, > Navin >
