Spark configurations when converted to usage mode automatically use the case insensitive map. All usages of options after should then be case insensitive.