| T.R | Title | User | Personal Name
 | Date | Lines | 
|---|
| 560.1 | it's automagic... | QUIVER::SLAWRENCE |  | Mon Dec 13 1993 12:23 | 34 | 
|  |     It's all automagic.
    
    The MAM queries each module to discover its interfaces and the signals
    each interface can use.
    
    When you connect a backplane port to a segment, the MAM finds a set of
    signals that can be used for the segment that interface can use.  If
    that search fails it checks to see if reconfiguring all interfaces on
    the backplane would result in all connections being made (all existing
    ones plus the one you just asked for); if so, it does the
    reconfiguration.
    
    So, in your examples:
    
    	If you connect the small module to the IMB first, that IMB would be
    allocated on the upper connector (the only possibility).  When you
    connect the large module, it would connect to the same signals; no
    problem.
    
        If you connect the large module to the segment first, it might be
    allocated on any flex channel.  When you attempt to connect the small
    module the MAM would figure out that only the upper channel would work
    and if the IMB was not already there it will order the large module
    interface to reconfigure.
    
    Note that if you were to attempt to connect a second small module to an
    IMB _different_ from an IMB that another small module is already on,
    the request will be rejected because even a total reconfiguration would
    not satisfy all connections (the first module would be undisturbed).
    
    Actually, the MAM will anticipate this situation if the hub contains at
    least one FDDI-capable module (including the bridge).  Since the FDDI
    modules can't use the flex channel on the upper connector, it will be
    allocated first for Ethernet segments because it is the least flexible.
 | 
| 560.2 | Thanks a lot  for the answer | ZUR01::SCHNEIDERR |  | Tue Dec 14 1993 08:08 | 0 | 
| 560.3 | So the Hub stop for a while? | OSAV20::IZUTANI | Kenji Izutani,Tech.Consul.,CSEC,DEC-Japan | Wed Dec 15 1993 18:27 | 9 | 
|  | RE:.1
So I can't stop thinking of the time duration for the backplane to 
reconfigure.
How long will it take to reconfigure the backplane? And will the hub stop 
working for a moment?
Kenji Izutani
 | 
| 560.4 | Your mileage may vary ;-) | QUIVER::GALLAGHER |  | Wed Dec 15 1993 19:48 | 13 | 
|  | >So I can't stop thinking of the time duration for the backplane to 
>reconfigure.
>
>How long will it take to reconfigure the backplane? And will the hub stop 
>working for a moment?
We haven't quantified "how quick" yet.  In general, the answer is "very quick".
So quick, in fact, that nobody has really worried too much about it.  
How quick is it for a person to swap cables in order to do a LAN 
reconfiguration?  Hub reconfiguration will be *much* faster than manual
reconfiguration.
						-Shawn
 | 
| 560.5 |  | QUIVER::SLAWRENCE |  | Thu Dec 16 1993 08:27 | 10 | 
|  |     I've gotten this question enough times now that I suppose I'll have to
    go set up an experiment to measure it.
    
    Nothing is disturbed on the backplane until the solution is already
    known, and then the operations all take place very quickly.  We have a
    monitor board that displays activity on a flex channel with a LED and
    the reconfiguration shows as a very brief flicker (some fraction of a
    second, I would guess).  
    
    Watch this space...
 |