Analysis of the Join Performance in Vertically Distributed Cloud Databases

Approach

The SeDiCo approach follows the principle of vertically distributed database tables. The main idea is to divide crucial database data and distribute them across different (public and private) cloud providers. Thus, every provider only gets a small part of the data. These individual small parts are worthless without the other parts, so enterprises are able to meet their compliance rules concerning data security and protection and so Cloud Computing becomes an interesting alternative to store vast amounts of data.

The main idea is to divide a basic database table into 2 partitions and to store these partitions at different cloud providers. Figure 1 illustrates the entire approach with Customer_Partition1 “p1” (in a public cloud) and Customer_Partition2 “p2” (in a private cloud). If now the data of one partition is stolen (e.g. creditCardNo of p2) this data cannot be misused, as the corresponding data (e.g. “name” and “dateOfBirth” of p1) is stored somewhere else. Moreover, only the owner of the data knows where.

Figure 1.

SeDiCo Architecture

Furthermore, as data are distributed across several cloud providers, their different programming interfaces - application programing interfaces (APIs) - become a major challenge. Different APIs have to be encapsulated in order to have a unique way of accessing the different interfaces. This encapsulation was realized with the jclouds¹ framework because of its Java support. Further alternatives are mentioned later the related work section.