Analysis of cache implementations in a vertically distributed cloud data store

This work is based on the SeDiCo framework (A Secure and Distributed Cloud Data Store) that implements a vertical database partitioning approach and distributes the respective partitions across different cloud providers. Unfortunately, the vertical data distribution suffers from severe performance issues, as the join of the data is a very time-consuming task. In order to overcome these performance challenges, we propose the introduction of a cache that stores frequently used data already joined in its cache memory. Although the introduction of a cache raises several new challenges (i.e. cache synchronization, replacement, loading strategies, etc.) it is a promising way to increase the performance of our framework. Thus, we focus on 3 concrete client-based cache implementations, namely an In-Memory key-value store, a locally installed database and a file-based JSON cache. The respective cache implementations offer appealing features for our framework and therefore we evaluate them in this work. We believe that the most important criterion for a cache is its performance and therefore we contrast the query performance results of each cache implementation as the main result of this work.