Hi all, we initiated a PR #803 <https://github.com/apache/systemml/pull/803/files> , to discuss the docker support for systemml. The following are the *options:* 1. systemml + pyspark + pip libraries *[Implemented in PR <https://github.com/apache/systemml/pull/803/files>]* - This is useful from the user perspective 2. build base OS image [CPU] + clone git repo + install all the spark, scala, mvn support. Example link here <https://github.com/tensorflow/models/blob/master/official/Dockerfile.cpu> 3. build base OS image [GPU] + clone git repo + install all the spark, scala, mvn support. Example link here <https://github.com/tensorflow/models/blob/master/official/Dockerfile.gpu> 4. Which image is good for deploying to kubernetes directly?, without the need for ambari and yarn.
First docker support is mainly focused for developers, so that we could test it ourselves. Let's keep this discussion on for a month. Regards, Janardhan
