Use Spark ML Pipeline on a REST webservice

Question

I have built up a Spark ML pipeline using a Logistic regression model, now I want to use this pipeline to do predictions on text through a REST webservice, what is the best way to achieve this.

Answer 1

Simplest open source approach would be using Apache Livy with its %json magic. A Scala or Python API with access to a spark session would be easiest custom solution. Azure ML would be easiest vendor specific option.