Conference netcad::hub_mgnt

You seem to be asking 2 questions here.  Yes, the DETMM allows each repeater
port to have up to 2 addresses. This is used so that you can set the
repeater ports to have security such that only a particular address will
receive the messages intended for it.  All other ports will basically
recieve a garbled message.  In addition, when security is enabled, it also
will only allow that particular address to broadcast from that port.

The second part of your message asks about port switching capabilities in
groups.  The DETMM is not capable of having different ports or groups of
ports on different LAN's.  That capability is planned for future products,
but will not be offered on the DETMM nor will the DETMM be upgradeable to
include this functionality.  

Cathy

    The current DECrepeater 900TM does not support grouping ports; packets
    are repeated on all ports (see below for an exception).
    
    There are a number of repeaters under development that will allow
    grouping ports in various ways and assigning the groups to different
    backplane LANs (or isolating them within the repeater).
    
    > This keeps the Repeater from being all consumed in repeating and
    > becoming a bottle neck
    
    The repeater can support the full media speed; by grouping the ports
    you are making the LAN smaller.   To make such smaller LANs really
    usefull you have to interconnect them with some higher-level device (a
    bridge or router).
    
    The exception I mentioned is the security feature of the 900TM; while
    this does not get you any extra bandwidth, it does address one of the
    other reasons why some customers will be interested in grouping ports. 
    A port can be configured such that only packets addressed to the device
    on the port will actually be readable from the port (unicast to its own
    address and all multicast or broadcast packets).  Unicast packets
    destined to other addresses will be garbled, preventing eavesdropping
    by normal devices that are in promiscuous mode. 

    When I mentioned grouping ports I'm not trying to move them to different 
    LAN segments.  
    
    Other vendors group either 4, 8 or 16 and address the group.  The ideal 
    case would be to individually address each port so when its repeating
    that is NOT repeating on unnecessary ports, therefore, its does not
    comsume the Repeater.
    
    	regards,
    
    	 kam

    RE: .3
    
    You mentioned in both .0 & .3 that there is a belief around that there
    are repeaters being produced which can not pass the maximum amount od
    data thrown at it, hence, causing the repeater to "become consumed and
    be a bottleneck".
    
    This sounds like great FUD being used by another vendor to sell his
    repeater product.  It is inconceivable that any vendor would make a
    repeater that could not handle the full data rate of an ethernet.
    
    If some vendor is, the product is garbage.  But, they product probably
    will pass the full load.  They are just using the ignorance of the
    distributor to try and build a case to sell their product, maybe
    because of another weakness that has yet to be seen.
    
    Mark
    

The only time we have seen any bottlenecking on repeaters has been on the
competition's product where they use a retiming module in a hub to do the
repeating for all the repeater modules in that hub.  The DECrepeaters are
each an individual repeater and do not have this bottlenecking problem.  In
fact, that is a feature of our product that you can use to sell against 
some of the competition that still uses the individual retiming module, not
the other way around.

    This new FUD 'save repeater cycles by switching them off' made my day.
    
    Side question: what is the repeater delay in our DECrp 900 ? I've always
    assumed/thought that Ethernet repeaters repeat signals within 1bit (100ns)
    or half a bit time (50ns). Am I correct ?
    
    Thanks in advance.

    Repeaters have a start packet propogation delay of between 3 and 7.5 
    bit times.
    

    Thanks for your response. I think I was mixing up with the original IBM
    Token Ring proposal. Do you know their current (token) repeat delay ?