We don't use a database. We store data in amazon AWS s3, and ElasticSearch, nowhere else. Spark isn't web service, Apache Spark. That's Spark Java which we don't use. We use kovert with vert.x instead.