Protection and Control of Information Sharing in Multics

From: Manish Mittal (manishm_at_microsoft.com)
Date: Fri Jan 09 2004 - 16:22:22 PST

  • Next message: shearerje_at_comcast.net: "Jim Shearers review of Information Sharing in Multics"

    This paper from Jerome Saltzer describes the design principles and the
    various mechanisms used for the protection and control of information in
    the Multics system. The key mechanisms described include access control
    lists, hierarchical control of access specification, authentication of
    users and memory protection. Paper ends with a discussion of certain
    known weaknesses in the current system.

     

    I found the five design priciples for protection of information really
    interesting. Even in today's system, these principles are not followed
    properly. Most of the current systems cannot run under least
    priviledges. Also, we tend to keep design secret from the attackers and
    fear that once it is known, system would be hacked in no time.
    Protection mechanisms should not be coupled with protection of keys.
    These principles are the basic premises for security. It is really
    surprising to see emphasis on these same principles for present systems.

     

    The paper then describes the primary memory protection . The multics
    virtual memory is segmented to permit sharing of objects in the virtual
    memory and to simplify address space management for the programmer.
    Addressing to the segments are done using descriptors. Protection
    information in Multics is associated with this descriptors rather than
    with the data itself. I found the concept of segment table, dynamic
    linking and descriptors quite interesting .

     

    The 'Authentication of users' in Multics is very interesting. Some of
    the features such as proxy login, timeouts, one way encryption reminds
    me of the kerb/digest protocol. Password protection techniques are also
    very noteworthy. Several logging and penetration detection techniques
    described in this paper are very similar to what we use in our
    Authentication system today. Logging incorrect login attempts, locking
    out accounts after 10 retries, introducing delays to frustate hackers,
    time/location of previous login attempts shown during the next login and
    monthly login reports are some of the features that are still very much
    prevalent.

     

    The author then describes several weaknesses of the system as perceived
    by him. This includes weak IO communication links, poor operator
    interface, weak passwords, supervisor interface and storage residues

     

    Overall, this paper is very interesting read. Author has described
    principles and mechanisms with good explanation and examples. Multics
    system is designed from security standpoint. Current systems tends to
    give precedence to usabilty & performance over security. Multics is an
    extreme example of an attempt at sharing while maintaining moduler
    flexibility. It ends up being too

    complicated, but it is an important example to study.

     

     


  • Next message: shearerje_at_comcast.net: "Jim Shearers review of Information Sharing in Multics"

    This archive was generated by hypermail 2.1.6 : Fri Jan 09 2004 - 16:22:29 PST