NETWORK OPERATIONS MANAGEMENT

There are a number of unique challenges involved in providing the communications infrastructure support for a nationwide backbone network. A data communications network is inherently a distributed system spanning a geographic area that may range from a room or an office to the entire globe. The Internet in particular poses a difficult environment in which to operate a networking service capable of providing production-quality support to ER research activities. (The world-wide DECnet poses a number of similar problems, and a few unique ones of its own.) The Internet is essentially an interconnected, distributed system of autonomous networks that are also distributed systems. Each of these autonomous networks is termed an autonomous system (AS), as it interconnects with other ASs in both an ad hoc and a sovereign manner. Production operation of the Internet therefore depends on the successful interoperation of an interconnected collection of such autonomous systems. A failure or an independent action in one AS can easily have an unanticipated impact on others.

In this environment, maintaining production-quality service requires full-time monitoring and problem resolution, and ESnet is operated 24 hours a day, 7 days a week. The ESnet Network Operations Center (NOC), shown in Figure 6-4, provides 24 hour-per-day monitoring and control capabilities for the various network components that comprise ESnet. The Operations Center is operated by the ESnet staff and NERSC supercomputer operations personnel. The NOC staff continuously monitors ESnet's backbone facilities to verify the network's integrity and to gather routine statistics that will facilitate troubleshooting and long-term planning. Electronic mailboxes for network information, network operations, and trouble calls are provided. An on-line trouble ticket system exists to ensure that all reported problems will be properly tracked. Requests for information about ESnet can be directed to info@es.net. Problems can be reported to trouble@es.net.

Figure 6-4. The ESnet Network Operations Center

Maintaining high network availability depends not only on problem resolution but also on constant planning, proactive implementation, and the management of critical transitions in carefully planned stages. For example, a number of schemes that attempt to define a next-generation Internet Protocol (IPng) were being defined and evaluated as this Program Plan went to press. However, the time scale required for such a major transition made a shorter-term solution essential. An interim proposal called Classless Inter-Domain Routing (CIDR) aimed to ease the demand on router resources by aggregating networks into address blocks. As of this writing, ESnet had just completed a major transition to an inter-AS routing capability called BGP-4 (Border Gateway Protocol, version 4), which is capable of supporting the aggregated addressing associated with CIDR by "tagging" network addresses. The implementation of these protocols was a major change that could not have been done piecemeal. Following several months of planning, a group of ESnet people totaling as many as eight at any one time completed the transition to BGP-4 in a single all-night session, with only minimal impact on users.

Network Infrastructure Services

Information Services

Videoconferencing Services