Remember that, although Spark is used for the speed of its in-memory distributed processing, it doesn't provide storage. You can use the Host (local) filesystem to read and write your data, but if your data volumes are big enough to be described as big data, then it makes sense to use a cloud-based distributed storage system such as OpenStack Swift Object Storage, which can be found in many cloud environments and can also be installed in private data centers.
In case very high I/O is needed, HDFS would also be an option. More information on HDFS can be found here: http://hadoop.apache.org/docs/current/hadoop-project-dist/hadoop-hdfs/HdfsDesign.html.
