We do have dedicated categories for Machine Learning. I would like to think that Spark ML subsumes those workloads, and using Spark ML, one builds machine learning applications.
Inasmuch as its broadness, we wanted to expand the summit to include AI use cases. As you know AI -> ML -> DL in the broad sense. And the categories allow submissions along those topics.
The startup time of a HDInsight cluster is quite long; it's not really suited for ad hoc clusters (which are really easy to spin up quickly in Databricks).
It takes around 20 minutes to spin up HDInsight clusters. So it is not instant but I wonder if databricks on Azure can do any better given they are running in customer environments so wont have a big pool of preallocated multitenant machines available.
AFAIK it is not multitenant, you get your own machines.
They spin up the VMs with either open source Spark or the Databricks runtime - you get to choose the distribution and version before spinning up the cluster.
agree, if you have batch jobs for spark, Azure Batch Service may not be a bad solution either.
HDInsight includes too much other complexity if all you wanted was spark.
Databricks will be interesting, as they can take away even they mysticism of touching azure beyond initially provisioning them some rights.
