[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: Agent Capabilities and sysORTable




>>>>> Thomas, Steven C (Steve) writes:

Steve> My question is: Why don't most agents (I haven't found any so
Steve> far) implement or at least populate the sysORTable?  Is there
Steve> some hidden difficulty in implementing it on the agent side?
Steve> Or have I misunderstood the purpose of sysORTable?

I personal answer to this question (which might of course be wrong) is
that the concept behind the sysORTable is unrealistic. The modelled
process works as follows:

(1) MIB designer writes conformance statements.
(2) Agent implementors document implementation capabilities relative
    to certain conformance statements in an agent capabilities
    statement.
(3) Agent implementors populate the sysORTable with pointers to these
    agent capabilities statements.
(4) Management applications read the sysORTable to obtain the OID
    pointers to load the relevant vendor specific agent capability
    definitions.
(5) Management applications can optimize their behaviour based on the
    agent capabilities.

Many implementations do not have formal agent capabilities statements
and hence there is nothing to refer to. So what people do is either
(a) not populate the sysORTable at all, 
(b) populate sysORTable for some other purpose (e.g. listing MIB
    modules supported in the sysORDescr object) without giving
    meaningful value to sysORID (usually 0.0),
(c) populate sysORTable by using the sysORID to refer to a 
    module compliance statement instead of an agent capabilities
    statement.

Even if an implementation does it all correctly as specified, I still
believe that a management application has a hard time to get hold of
the agent capabilities definition (step (4) above) so that it can take
advantage of it. There is also a non-trivial risk that the agent
capabilities definition is actually not in sync with the
implementation. In order to be on the safe side, the manager must be
able to recover gracefully anyway if something does not exist or can't
be changed as specified...

You may also want to read Dave's Question Answered column in the
Simple Times 8(1) published in September 2000 which addresses this
topic.

/js

-- 
Juergen Schoenwaelder      Technical University Braunschweig
<schoenw@ibr.cs.tu-bs.de>  Dept. Operating Systems & Computer Networks
Phone: +49 531 391 3289    Bueltenweg 74/75, 38106 Braunschweig, Germany
Fax:   +49 531 391 5936    <URL:http://www.ibr.cs.tu-bs.de/~schoenw/>