Data splitting makes a local partition of the sensitive data and separately stores data fragments in different CSPs, in a way that each individual fragment does not cause privacy risks; data fragments are stored in the clear without any modification; hence, the data accuracy and the analytical interest are efficiently preserved. Since data are split and only CLARUS knows the exact cloud locations for a given dataset, CLARUS performs an adaptation (and/or orchestration) of user queries and an aggregation of the results. 

(Efficiently*) Supported operations: 
Storage
Search
Retrieval
Update
Computation
Performance impact on local premises (per data size): 
Constant for all operations (including storage)
Data accuracy preservation: 
Full on the partial data stored by each CSPs
Full for CLARUS users
Access of non-CLARUS users: 
NO
Security: 
High if CSPs are independent to each other
Management: 
Several CSPs or CPS accounts are needed
Operations are transparent for the end user and the CSP
Some computations require specific modules on the CSP