January 15, 1998

PLACE:                  EP 309
TIME:                     2:30 PM
ATTENDEES:      Professor Junk
                               Don Tobin
                               Zac Woodall
                               James Cooper
                               Jamie Marconi
                               Sam Gerlach
TOPIC:                  Prototype

AGENDA:

Brief Introduction
 ------------------

 (taken from the existing CS481 notebook)
 
The purpose of HummingBird is to gather security information off
of the network and pass this information to other hosts.  Examples
of information include logfiles, load, and number of users.

Within an hierarchy, superiors give directives and subordinates
pass desired information up.  At the moment, passing security
information protocols is the biggest stumbling block.

The actual Hummer is the decision maker.  It decides what is kept
and/or logged.

The current task is to recreate the existing HummingBird in some
ways.  An example of subordinate management was displayed in a
 graph.

Requirements, Limitations and Constraints
 -----------------------------------------

The final product will be run on HP and Solaris systems running
 HP-UX 10.10 or 10.20.  The existing code will be built on, with
 the current format preserved.

Priority One:

 - Add peer-to-peer management (with a certain level of trust, as
                noted by Don Tobin)
    This includes allowing individual hierarchies communicate
    with each other.
 - Information Flow Policies
    We need to decide between straight or broadcast
    communication.  Also, decide to what degree and how
    hummers should cooperate.
 - Autoregression Testing
    We need to automate the existing test suite for
    developmental testing and implement operational
    testing.

Priority Two:

 - Autoconfiguration System
    Currently you need to go to each host individually.  This
    can take 30 minutes plus.  In the future, we want to make
    this more top-level.  Ideally, the manager will provide
    the hierarchy and have it pushed down the tree.  This may
    not be faster, but it will be easier.

- Integrate ATOM (by Dean Polla)
   This tool allows you to configure groups with threat
    levels.  We want to add this to the manager/subordinate
    configuration and push it down the hierarchy.

 - Improve/handle data integrity and trust issues (Don Tobin
     suggested moving this to Priority One)
 

Priority Three:

 - Paging service
 
 - Update visualization system

 
Solutions
-------

Priority One:

- We plan on implementing the peer-to-peer management as a type
    of listserve.  A manager will be assigned, from the peer
    group, to oversee the communication.  The manager will
    handle requests and route messages.  Some type of protocol
    would have to be set up the add members (broadcast,
    vote...)
 
   (Don asked about the problem of a manager failing, which
                we have not yet addressed)

- Information flow will be handled with a broadcast type of
    service.  The clients will send information to the
    manager and the manager will pass the information on.  A
    cooperation protocol will be defined similar to the way
    people cooperate in an e-mail system.

- Both white and black box testing will be implemented.  They will
    be automated into a test suite.

Priority Two:

- The autoconfiguration system will entail evaluating the current
    configuration method and making changes to ease the
    process.  There are three current  helper programs to
    HummingBird:  Apache, Kerberos, and Postgresql.  This is
    where the majority of configuration comes in.  However,
    these helper programs are needed to keep HummingBird less
    complicated.  This is our dilemma.

- Dean's project will be integrated into the current HummingBird.

- (moved to Priority One) Various methods will be used to increase
    data integrity.

Priority Three:

- These are optional features and have not been explored in depth
    at this time.
 

Directives from Professor Junk
------------------------------

Team Meetings Tuesday and Thursday at 2:30-3:00 in Professor Junk's
office.

TODO:

(By 2:30 January 20, 1998)

- Vision Statement
        grand view
- Project Description
        100-150 words
        suitable for a web page
- Team values
        dividing chores
        commitment
        responsibilities, etc...
- Risk Assessment (Top Ten List)
- Team leader approach
- High-level schedule
        Gantt chart
        concentrate on first month
        milestones
        task list
- Read Syllabus
- Features/Requirements list
        low-level
        functionality
- What first prototype will be able to do