USATLAS Network Performance Monitoring with perfSONAR Netcast Archive
We will present the deployment and usage of perfSONAR in USATLAS, the United States collaboration in the Atlas experiment at the LHC (Large Hadron Colider) at CERN (European Organization for Nuclear Research).<br><br>
The Atlas computing infrastructure is organized in Tiers of sites with distributed data storage and computing queues. At the national level one US Tier 1 center and five geographically distributed Tier 2 centers support collaboration-wide computing, while many institutes are building local Tier 3 centers with additional private resources for fast turnaround on local analysis efforts. High speed networking is a crucial enabler of this distributed computing model, and USATLAS has successfully deployed perfSONAR to instrument all its Tier 1 and Tier 2 interconnections, with Nagios adding a system-wide centralized status dashboard. Efforts are ongoing to extend this monitoring coverage to all US Tier 3 sites. We will illustrate the presentation with some examples where our perfSONAR installation has assisted in the detection and diagnosis of networking problems.