A report on the effect of heterogeneity of the grid environment on a grid job

Abstract

Grids include heterogeneous resources, which are based on different hardware and software architectures or components. In correspondence with this diversity of the infrastructure, the execution time of any single job, as well as the total grid performance can both be affected substantially, which can be demonstrated by measurements. In need to effectively explore this issue, we decided to apply micro benchmarking tools on a subset of sites on the EGEE infrastructure, employing in particular the lmbench suite, for it includes latency, bandwidth and timing measurements. Furthermore we retrieved and report information about sites characteristics, such as kernel version, middleware, memory size, cpu threads and more. Our preliminary conclusion is that any typical grid can largely benefit from even trivial resource characterization and match-making techniques, if we take advantage of this information upon job scheduling. These metrics, which in this case were taken from the South Eastern Europe VO of EGEE, can provide a handle to compare and select the more suitable site(s), so that we can drive the grid towards maximum capacity and optimal performance. © 2008 Springer.