GrayWulf: Scalable Clustered Architecture for Data Intensive Computing

  • Alex S. Szalay ,
  • ,
  • Jad Vandenberg ,
  • Alainna Wonders ,
  • Randal Burns ,
  • ,
  • Jim Heasley ,
  • Tony Hey ,
  • Maria Nieto-SantiSteban ,
  • Ani Thakar ,
  • Catharine van Ingen ,
  • Richard Wilton

MSR-TR-2008-187 |

Pre-Print of version submitted to HICSS 2009.

Data intensive computing presents a significant challenge for traditional supercomputing architectures that maximize FLOPS since CPU speed has surpassed IO capabilities of HPC systems and BeoWulf clusters. We present the architecture for a three tier commodity component cluster designed for a range of data intensive computations operating on petascale data sets named GrayWulf†. The design goal is a balanced system in terms of IO performance and memory size, according to Amdahl’s Laws. The hardware currently installed at JHU exceeds one petabyte of storage and has 0.5 bytes/sec of I/O and 1 byte of memory for each CPU cycle. The GrayWulf provides almost an order of magnitude better balance than existing systems. The paper covers its architecture and reference applications. The software design is presented in a companion paper. † The GrayWulf name pays tribute to Jim Gray who has been actively involved in the design principles.