DISCOA: architectural adaptations for security and QoS

  • Authors:

  • Omer Erdem Demir;Prem Devanbu;Nenad Medvidovic;Eric Wohlstadter

  • Affiliations:

  • Computer Science, UC Davis, CA;Computer Science, UC Davis, CA;University of Southern, California, Los Angeles, CA;University of British Columbia, Vancouver, B.C., Canada

  • Venue:
  • SESS '05 Proceedings of the 2005 workshop on Software engineering for secure systems—building trustworthy applications
  • Year:
  • 2005

Quantified Score

Hi-index 0.00

Visualization

Abstract

Modern distributed systems have greatly benefited from developments such as model-driven development, and architectural description languages. Abstract models of components (e.g., IDL) and models of interconnection (e.g., architectural description languages, or ADLs) provide important software engineering advantages, such as explicit design models, type-checked integration across machine and language boundaries (with generated marshaling and dispatch code), the possibility of third-party components, and automated verification of design artifacts. But, when distributed systems are enhanced to provide security features, many of these advantages do not apply. Security features are hand-written into almost every part of the system; there is no explicit component or architectural model, or separable "security component" security code fragments are scattered and tangled through the different distributed elements of the system, and are often reduced to communicating through lowest-common denominator fragments (like raw bytes) since they are not represented in the model.In this paper, we describe DISCOA, a proposed extension of our earlier work on DADO [23] to handle security features in distributed systems, using explicit architectural models with aspect-oriented extensions.