Managing File Subsystem Data Streams for Databases on Networked Systems

Abstract

One important activity for networked database systems that distribute data across several workstations is moving data between the file and network subsystems. It is possible to create data streams in the operating system kernel. If provided on a system, they allow user level processes to request transfer of data without having t copy it into the user space. This is particularly useful for data whose content or format is not modified during the transfer. In this paper we present a conservative criterion for access and control for the management of such data streams for databases in a networked environment, and define the implementation requirements for achieving the criterion. The approach is to maintain at least the current level of access management. We define the specific implementation semantics that this criterion entails.