Fundamental challenges for new DataCenter

Fundamental challenges for new DataCenter
Ronald Luijten
Manager Data Center Optimization, IBM Research - Zurich

DEI - Sala seminari
25 maggio 2011
Ore 16.30

Abstract:

I will introduce you to more's law (yes, with 1 o), and present 5 walls that imply fundamental changes to future DataCenter architecture. This is compounded with new Data Center workloads and technology disruptions. Cost and performance optimization criteria will be on energy and data motion rather than clock frequency and computing. I will make clear that the architecture will be significantly different however we don't know how to synthesize it, and discuss a new design exploration tool we are developing to help us find good implementations.

Short Bio:
Ronald P. Luijten joined IBM zurich research in 1984 and manages the data center optimization team. Ronald’s research interests are in datacenter I/O and interconnect architecture (‘Data Motion in Data Centers’) and microservers. His team is also working on 2.5D pattern classification for developing 22nm CMOS design methodology. He holds more than 25 issued patents, two outstanding technical achievement awards and a corporate patent award.

Contatti:
Giuseppe Serazzi

Area di ricerca:
Architetture