Data Protection and Security

V

Access Control

V.II

Unix and Windows NT Access Control Overview

At an operating system, access control is generally modelled by a matrix of access operations, with columns for objects and rows for subject. This is depicted in the table below.

In this table, there are four types of access operations: read, write, execute and append (sometimes also referred as blind write). Note that the total number of access operations can vary in different implementations. On the most elementary level, there should be two access modes (observe and alter) and most other access operations could be expressed in terms of these.

This access control matrix is typically a conceptual vehicle, not an implementation structure. For implementation there are two obvious options. Access control lists (ACLs) are the columns of the access control matrix and capabilities are the rows of the access control matrix. Using either of these, only the nonempty elements are stored and therefore disk space is not wasted.

Example:

ACL for /mail/ali: Ali: r,w sendmail: a

Bora's capability: a.exe: r,w,x sendmail: r,x

Operating systems like Unix and Windows use ACLs as their basis for access control. The second option, capabilities have not become a widely used security mechanism however the increase in use of distributed applications has made capabilities again a viable alternative. This is because roaming users are easier to handle with capabilities.

V.II.I Q