11-28-2011 05:53 PM
100 GbE is going mainstream, and that was no more apparent than at SuperComputing 2011 in Seattle this year.
Today’s cutting-edge research has produced exponential growth in the volume of scientific data generated by research projects, numerical simulations, compute clusters, and machine-to-machine interactions. Accommodating this rapid data growth is a key challenge for scientists who manage and transfer data sets, as well as researchers who participate in widely distributed collaborations and require reliable, secure access to the data.
Indiana University demonstrated this in its Data Superconductor at the SC11 SCinet Research Sandbox (see press release). The Data Superconductor is a Lustre-based, high-performance file system optimized for operations over a WAN. A complete cluster and file system operated at each end of a 100 GbE link powered by Brocade MLXe Routers, spanning approximately 2300 miles between Indianapolis and Seattle. Within each compute cluster, Brocade 1860 Fabric Adapters simultaneously connected the servers over 8 Gbps Fibre Channel to the SAN storage devices and 10 GbE to Brocade VDX switches.
This demonstration showed how scientists can collaborate across administrative domains using multisite workflows and distribute data from instruments to compute resources. With a centralized file system like this serving thousands of computational resources around the world, user data can be available everywhere, all the time. Hm.. this sounds very similar to a five letter word that seems to be on the mind of every CTO, network architect, IT manager, etc, etc..