When looking at the various legal issues relating to cloud computing, it is important to understand the technical details of the particular storage/service being considered.  While all may at some levels seem the same, working out the details of service level agreements, privacy and security provisions, and disaster recovery options requires knowing which of the several new computing paradigms are emerging from large commercial clouds are being used.

On variation that is prevalent involves virtual machine based utility computing environments such as Amazon AWS and Microsoft Azure. On a virtual machine, one must be aware of how and where data is transferred in the processing performed by the virtual machine.  Also, the particular potential security vulnerabilities should be addressed, as they are quite different from that of physically secured computing devices.  Indemnification may also be a difficult issue because of the ubiquitous nature of determining sources of error.

Another cloud computing variation involves new MapReduce programming paradigms coming from the Information retrieval field which have been shown to be effective for scientific data analysis.  In the MapReduce environment, problems are partitioned and de-aggregated by inputs and computation steps into a multitude of sub-problems. Eventually every sub-sub-sub-problem is resolved by a cloud resource, then the solutions are successively combined to create a final solution. As the MapReduce environment dynamically partitions and re-assembles, the actual locations and incidences of transmission are not know beforehand (and may be difficult to determine afterwards in event of an error). 

For any cloud computing project, the location, security and storage of the data may have legal significance, and the underlying cloud computing technology needs to be considered when framing the legal protections.  While a “cloud computing environment” may sound well defined, in many contexts additional disclosures are needed to see if a particular solution is appropriate for the nature of the data.