A Framework for Fast MapReduce Processing Considering Sensitive Data on Hybrid Clouds
Shun Kawamoto, Yoko Kamidoi, Shin’ichi Wakabayashi · 2020
In recent years, large-scale data needs to be analyzed and processed for various purposes. One of the methods for processing large-scale data is to use the cloud. However, general public clouds have security issues. Therefore, when handling confidential data, there is a method of processing with a hybrid cloud that consists of a public cloud and a private cloud which users can use safely. In order to use the two clouds efficiently and securely in this hybrid cloud, the distributed processing framework SEMROD (Secure and Efficient MapReduce Over Hybrid Clouds) was developed. However, SEMROD has two problems: communication load and deterioration of load distribution efficiency. In this paper, we propose a method called HFK-SEMROD (Hold First Key-SEMROD) to make the existing method SEMROD more flexible and faster. Moreover, we provide two additional options of the HFK-SEMROD, whose utilizations are selectable by job types. Finally, we compare its performance with SEMROD experimentally.