CNS-CSR- Understanding and modeling the trade-offs in data centers for next-generation sustainable management Data centers, as an IT infrastructure component, permeate all large-scale computing domains, including scientific computing and simulation, cloud computing, banking, online gaming and entertainment etc. In other terms, any organization above a certain size owns or rents a data center in some way. The cost of maintenance keeps rising despite the computing technologys increasing efficiency trends (performance efficiency or energy efficiency). A key approach to tackling the rising costs is energy-adaptive computing, which adapts the performance to the given energy constraints. Current commercial recipes, such as follow-the-moon or chiller-less data center are not guaranteed to work in all data centers, because of the inherent traits of those data centers they were demonstrated to work on. The reason behind many solutions not working is because there are several cyber-physical trade-offs (energy vs performance, time vs energy, cooling vs computing, communicating vs computing etc) only few of which are considered by each such recipe (for example, follow-the-moon utilizes the trade-off between spending energy to communicate vs spending less energy to cool while paying less per energy unit).
|Effective start/end date||8/1/12 → 7/31/17|
- National Science Foundation (NSF): $466,000.00