There are some commercial distributors of Impala (like MapR) that have presumably not used CDH.
I wish I had a better answer to your questions; I know a number of Impala contributors are considering this. I imagine the most prominent impediment is finding time to get it done. On Thu, Jan 12, 2017 at 6:04 AM, Joseph Naegele <[email protected]> wrote: > Hi folks, > > We have a custom, non-CDH Hadoop/Spark cluster on CentOS 7and would like to > evaluate Impala for our needs. What would be the easiest method of deploying > Impala in our situation? We're not prepared to migrate to CDH at this time > (is this a requirement?) > > 1. Build from source and install - this seems overly complicated right now > and requires a full build system, but achievable > > 2. Install from CDH repository - straightforward, but pulls in Hadoop, Hive, > ZooKeeper, etc., all of which we already have installed and customized. > > 3. Use CDH Docker image - this might be complicated since Impala needs to > communicate with HDFS, Hive metastore, etc. > > What I really want is a standalone binary distribution of Impala, but I > understand if this doesn't exist yet. Any suggestions? > > Thanks! > --- > Joe Naegele > Grier Forensics > >
