Behind the scenes of NCSA: Work for bandwidth | National Center for Supercomputing Applications at the University of Illinois
Behind the scenes of NCSA: Work for bandwidth
08.27.12 - Permalink
Have you ever wondered how your computer, internet phone service, laptop, smartphone, or gaming machine gets to the resources it needs to operate and provide you service? Or how NCSA supercomputers transfer data to sites thousands of miles away? It's the network!
What would it be like to come to work and not have a functioning network? The NCSA Network Engineering group ensures that doesn't happen. Team members Tim Boerner, Nick Buraglio, Paul Wefel, and David Wheeler see that NCSA staff, computing resources, and researchers are served by networks that are fast, reliable, and adaptable.
To accomplish this requires thinking of networking in many layers, like an onion. The layers start at the outside with the physical copper and fiber cabling and go all the way inside to operating system specifics and application performance tuning. A deep understanding of the details of every one of these layers and the interdependence between them is required. Building a resilient and adaptable network requires complex engineering and architecture plans, deep knowledge of routing and switching platforms and protocols, as well as the solid fundamental knowledge of all the layers.
Supporting services like Dynamic Host Configuration Protocol (DHCP) and Domain Name System (DNS) must work well for the network to appear functional. Often when a system is having a communications problem, people assume it's the network, so Net Eng must locate the true source of the problem. The team also monitors the many circuits that make up the NCSA LAN and WAN for errors, outages, and utilization, then uses the data to trend usage patterns and plan capacity upgrades.
And they work closely with many other NCSA groups, like collaborating with security to provide feeds for a monitoring cluster, helping track down problematic hosts, and implementing security policies; working with systems and storage to design both ethernet and infiniband networks that will scale yet provide fast, reliable communications for the supercomputers; and providing network solutions for research needs to the NCSA labs that would not be possible within the normal environment.
Much of what goes on at NCSA requires connectivity to external organizations. NCSA maintains a number of wide area circuits connecting the center to other networks like XSEDE, MREN, OmniPoP, ICCN, and backbone networks like Internet2 and ESnet. Net Eng maintains all of that connectivity, plus works with peers at other institutions and at the regional network providers to make sure NCSA is positioned for connectivity to the next generation of backbone networks.
So the next time you wonder why it takes a few seconds for a web page to open, consider instead how extraordinary it is to be able to access the world from your fingertips. Think about everything that has to happen in the background, on the network, to bring you that web page. Then thank a network engineer.