Conference 7.286::fddi

    
    The usual answer is, no, you only need to worry about the length of
    cable involved. In some special cases you can play with the numbers. 
    
    Do you have to do a custom design of some type? 
    

yes, I have a configuration to design for Eurocontrol (European Air Traffic 
Control).
They have very drastic performance requirements.

Typical requirements are, for example, 

1. What is the minimal value for TTRT and how to modify it ?

2. What kind of frames are transmitted and received by A port to M port 
(secondary) in a normal use (dual homing) ?

3. How to detect a problem in A port in a normal use and how can you monitor 
it in a normal use ?

4. What is the accurate mechanism for ring insertion in a SAS and DAS/Dual 
Homing configuration ?

I think they are afraid to use FDDI technology because they don't know it as 
well as Ethernet.



						Thanks,

						Richard

>>1. What is the minimal value for TTRT and how to modify it ?

    I believe the range is from 4ms to about 165ms.  It's not changed
    directly, but rather is decided through the claim process as each FDDI
    node on the ring requests a particular value, with the lowest bidder
    winning.  Digital's FDDI products ship with a default request value of
    8ms.  There is a paper by Raj Jain (formerly of Digital) that describes
    why this is a good value and why exceedingly large values can be
    detrimental.
    
>>I think they are afraid to use FDDI technology because they don't know it as 
>>well as Ethernet.

    I don't know the low level SMT protocol well enough to answer your
    other questions, but that in and of itself tells you that you DON'T
    need to be intimately familiar with FDDI to use it.  I'm been using it
    for over four years now and I still learn more about it every day.
    
    Yes, FDDI can be more challenging than Ethernet, but even Ethernet has
    its share of ramp up with the various network misconfigurations and
    problem detection that can and does occur on an Ethernet network.  FDDI
    has the flexibility of offering a very simple network design to start
    with, followed by more intricate network designs as network managers
    become more familiar with the technology.
    
       - Larry

    
    Here are some answers to your last questions-
    
    1) The minimum value of TTRT is generally assumed to be 4 ms, although 
    it can be as low as the value used for TVX, which is typically 2.5 ms. 
    You set this value by adjusting the value of T_Req, via the Requested
    TRT slider in Hubwatch. You can also set it via an SNMP NMS. There is 
    very little reason to play with this slider, unless you know EXACTLY 
    what you are doing. If there is a real-time response requirement for 
    this system, you will not control it simply by adjusting TTRT. Send me 
    some mail with all the details of how you propose to design the network
    if you think you will need to adjust these values. Just setting TTRT to 
    2.5 ms will not guarantee a 2.5 ms response time. 
    
    2) We need to get some terminology straightened out. Lets use the words
    "standby connection" to describe the A to M port connection of a dual
    homed station. The word "secondary" has a precise meaning in FDDI - it
    refers to the secondary ring, which is not part of this connection. 
    
    Now, when the A to M connection is in standby mode, there are no frames
    (e.g. LLC frames or SMT Frames) present on it. The connection continuously 
    runs the PCM connection sequence, using a 50 second duration for the Link 
    Confidence Test. At the end of the PCM sequence, the DAS station 
    determines that the configuration is Dual Homed, because it can see
    that it has both A to M and B to M connectiions being offered to it. 
    Since the dual homing rules require that the station use the B to M 
    connection when dual homed, it does not allow the A to M 
    connection to come up. If the B to M connection should fail, the DAS 
    station restarts the A to M connection sequence, using an LCT time of
    500 ms. At the completion of the PCM sequence, the connection is
    allowed to complete, because the B to M connection is not available. 
    If the B to M is later restored, the B to M will be restarted, so that
    the preferred topology is restored. 
    	
    3) If by normal use you mean when the link is in standby mode, the
    problem detection is the same as it is for any other new connection.
    The PCM sequence will determine if the connection is good enough to be
    allowed into the ring, and take the connection if the rules allow it. 
    
    4) I'm not sure what the question means, but I'll give this response,
    and we can fine tune it. 
    
    Before any connection is made to a ring, the two ports that will be
    connected run the PCM sequence. In this sequence they exchange details
    of their port types and run the Link Confidence Test. If the ports to
    be connected are allowed (e.g. M to M ould not be allowed), and the
    Link confidence test completes without errors, the stations connect the 
    ports together. After they are conneted together, the ring is scrubbed,
    which means that idles are sourced for a period of time guaranteed to
    strip all remeianing frames from the ring. This prevents olf frames
    from being present on a newly formed ring. After that, the all stations 
    in the ring bid for the token, using the claim process. the station
    that has the lowest requested TTRT issues the new token, and the ring
    is operational.