Performance Study of Coding Schemes

Abstract

The time for big data has come. Quintillions of data are created every day. The size of total data has caused problems with storage. Many different institutes and business store their data in the data centres with more than hundreds of racks containing storage nodes. But those storage nodes are hardware devices that can fail from time to time. As a result, data may be lost. Therefore, protection of the data is required and can be done by adding redundant information using coding methods. When one or more storage nodes are failed, a repair process is triggered to regenerate the lost data. In this paper, a multi-rack storage system is considered. Due to system architecture, the repair process can be classified into two type, namely, the Inside Rack Repair Process (IRRP) and Across Rack Repair Process (ARRP). Since these two types of repair process incur different costs to the system, a new coding method is needed to minimize the repair cost.