Spark is good if it fits your use case (or if you can make it fit your use case). We have definitely found that scaling can be a real issue though.