Development of Platform for Hard Disk Deive Failure Prediction

Abstract

In the operation of the Multifunctional Information and Computing Complex (MICC) of the Meshcheryakov Laboratory of Information Technologies (MLIT) at the Joint Institute for Nuclear Research (JINR), a large volume of server equipment is utilized, which provides computational resources to many scientific groups and experiments. Some components of this equipment (primary hard drives, as well as CPUs and memory modules) are subject to wear and must be replaced in a timely manner to ensure the uninterrupted operation of the computing complex. For the replacement of failing equipment, MICC maintains a spare parts warehouse that requires replenishment. This paper presents a system and tools for collecting, storing, and processing information about hard disk drives for future predictions of HDD failures and improving spare parts inventory planning. As an example, a statistical method based on Weibull distribution was used on a collected data to provide basic estimations of HDD failures depending on their working time. The developed solution is built on freely distributed components and can be used in similar infrastructures. Integration into the inventory system of the JINR Cloud infrastructure is also planned.