In cooperation with ACM SIGMOD/PODS and sponsored in part by AT&T Labs, we ran a workshop on network-related data management (NRDM) on May 25, 2001 in Santa Barbara, California. The following is a brief report on the motivations and outcome of the workshop.
Motivation: A significant amount of data is being steadily captured at various levels on the Internet. This includes both trace data and configuration data, and is being used increasingly to discern the topology of the Internet, to understand routing, and to characterize Internet traffic patterns. The NRDM 2001 workshop was intended to bring together researchers and practitioners in the database and the networking communities, to understand the challenges of managing and analyzing network related data.
Outcome: The workshop included an invited talk by Vern Paxson (ACIRI), eight refereed papers in three technical sessions, four mini-panel sessions, and a concluding discussion section. There were 30 participants from various industrial labs and academic institutions. Various problems involving the difficulties with large scale Internet measurements were highlighted, such as the ad hoc nature of longitudinal data (varying fields gathered over different periods of time, subtle details such as omission of outliers not recorded with the data), and the complexities of managing and sharing the data. Suggestions for fixing some of these problems were then discussed at length, including the use of technologies like data warehousing and data mining, metadata management, and scientific data management techniques like maintaining lineage and audit trails within the database.
Balachander Krishnamurthy, Divesh Srivastava.
