<p>The manufacturer of data center and the literature have pointed out that the performance of magnetic disk drives in the server will degrade due to various operation factors. For instance, when vibrations caused by noise and unbalanced flow field is close to resonance frequency of the server chassis structure, resonance will occur leading to degradation of disk drive performance.&nbsp; If ambient temperature is abnormally high, the disk drives may be damaged or the performance will be degraded. When the above problems occur, the amount of data stored in the global community software, smart grid or enterprises will be lost, which will be a very serious problem.</p><p>To fully understand the root causes of the mentioned problems, in this paper, the manufacturer's relevant experimental settings and disk drive’s data sheet are first examined. Despite of varieties of disk drives, linear and rotational vibrations experiments of a single disk drive are then conducted to explore the severity of the impact of external linear and angular acceleration on the performance of the disk drive. Secondly, the maximum sound pressure generated by the maximum speed of the server fan are measured to assist understanding of air-born and structure-born vibrations to the disk drive’s data access performance. Finally, the resonance frequencies of the whole server structure including chassis and disk drives are determined experimentally by modal tests. With fan speed set at its highest level, the acceleration gauge is used to measure the frequency corresponding to the maximum response of the chassis to confirm whether resonance happens and the whole response is amplified.&nbsp; It is found in our investigation that assuming that the resonance frequencies of the chassis, fan and disk drive are similar, it is very likely that resonance will occur and the overall responses will be amplified. In order to identify the above-mentioned problems and make effective classification based on available information, the theory of decision tree ID3 is proposed and examined in this paper to establish a preliminary model to make the classification interpretable.</p>

Identification of Factors Affecting Disk Drive’s Performance in Data Server by Use of  Decision Tree Learning Method

Paper Proceedings: https://asmedigitalcollection.asme.org/ISPS/ISPS2021/volume/84799

ASME 2021 Information Storage and Processing Systems (ISPS2021)

Supporting the need among researchers worldwide for additional online and multimedia content, ASME now offers the ASME Conference Video Collection. This platform will collect and provide long-term access to conference presentation videos, encompassing all technology areas within ASME's technical conference and events program. About 30,000-40,000 engineers across all mechanical engineering and related disciplines participate in ASME conferences each year. This includes authors of approximately 6,000 to 8,000 technical papers published by ASME, the majority of whom also submit video presentations for our virtual conferences. The video collection will be available as an add-on to existing product packages or as a separate subscription.