Spark is a way of processing data, ideally stored in a system of record (Hive/HDFS/S3/MemSQL etc).
They're not the same.
For the kinds of processing both Spark and MemSQL do (e.g. join operation) is Spark faster than MemSQL?