Performance evaluation and modeling of active tile in raised-floor data centers: An empirical study on the single tile case

Raised-floor data centers usually suffer from the local hotspots resulted uneven cool air delivery. These not only degrade server performance, but also threat equipment reliability. The commonly used industrial practice of increasing Computer Room Air Conditioner (CRAC) blower speed for removing is energy inefficient and may lead to overcooling some servers. In this paper, we explore potential active tiles in center cooling management. particular, deploy a prototype tile production conduct extensive experiments investigate performance. It shown that deploying with even 10% fan increases flow by 49%, sealing under-rack gap reduces rack bottom temperature up 6°C. Moreover, three machine learning techniques, i.e., Gaussian Process Regression (GPR), Artificial Neural Network (ANN), Multivariate Linear (MLR) are employed construct end-to-end data-driven thermal models tile. Using field measured as training testing sets, it concluded GPR ANN competent accurate modeling tiles. Specifically, achieves smallest prediction error which around 0.3°C.