It takes about as long as it takes to start up a Spark cluster, and you can interact with it entirely through Spark APIs if that's what you're comfortable with. It can also be used in "Split Cluster Mode" so you can use your existing Spark build instead of what's embedded within SnappyData if you prefer.