The HPC Cloud has one large storage pool built with Ceph. 900TB are available. You can make images (that will look as hard drives to your VM) on that shared storage for large quantities of data (think of data larger than 100GB). This should therefore be the storage option for working with data, in a similar fashion as the "Scratch" space is used in other computing systems. Ceph is a scalable system that provides great benefits in a system such as the HPC Cloud, where many tenants (users) may be making I/O calls concurrently for long periods of time. Our Ceph implementation is specially good for applications which require lots of writing to disk, because we use caching in SSD disks that will prevent lots of IO from stalling while waiting to be written to disk.
For installing and running an operating system, each compute node in the HPC Cloud is also equipped with local storage on SSD drives. If you keep your OS images small, they can fit in the local SSD disks and your VM starts and shuts down faster as well.