Conference 7.286::fddi

>o	For Dual-Homing do the interfaces on the Systems have to be DAS or is
>there such a thing as a SAS with Dual-pairs (one for redundancy in Dual-Homing)

Dual homing exists only for dual-attached things (concentrator or station).

You can get a somewhat similar effect by using two separate SAS adapters,
but that's not the same thing.  A closer analogy for that approach is an
Ethernet with two adapters connected to the same cable.  It also gives you
redundancy.  But it means you need to have higher layer protocols that allow
multiple LAN adapters connecting to the same LAN.  (For example, DECnet Phase IV,
or DECnet/OSI with Phase IV addressing enabled on both ports, does not allow
doing this.)

>o	If you do need a DAS interface for Dual-Homing, I understand that one
>pair goes into a backup mode. Does this make it act as a SAS in the sense that
>like a SAS it is not part of the ring or is it acting like a DAS with one ring
>being unavailable. I ask this as it makes a difference as to whether you can
>remove workstations/ systems without disturbing the ring.

Your description of how SAS and DAS act is a bit confused.  Obviously, a station
that's connected to the FDDI is part of the ring...

A DAS with one port in backup (or disconnected, for that matter) acts a lot
like a SAS, but it is not technically correct to call it a SAS.  (Some people
do this, but that is not proper.)

Any station connected to a concentrator can be removed and reconnected without
affecting any other station, other than a very brief (fraction of a second)
interruption in network service when the change is made.  In particular,
unlike stations connected to the trunk dual ring, stations connected to
concentrators can be removed without risk of network partitioning.  This is
why it's a good idea to use concentrators...

>o	With this kind of requirement are there any alternatives to provide a
>redundant FDDI backbone?
>
>The two DEChub 900's have to be connected together to form the
>redundancy within the ring. A few questions arise from this. I would like 
>to connect the A port of one concentrator on one DEChub to the B port of a 
>concentrator in the other DEChub. Additionally I need to connect the 
>reverse pair via two other concentrators in the hubs. This would avoid the
>situation where - if only two concentrators were involved rather than 4 - the
>failure of one concentrator would partition the two hubs.

Could you draw a picture of what you mean, what kind of faults you're
talking about?  I can't visualize this...

>o	Is it a valid configuration to connect the concentrators of one hub to
>the concentrators on another hub in the fashion mentioned above or do I need 
>to run FDDI bridges between the hubs?

There's nothing special about the hub.  Think of it as a fancy bracket that
holds the concentrator and supplies it with power.

>o	Is it valid to take the A port from Concentrator-1 in Hub-1 to port B
>Concentrator-2 Hub-2 and port A Concentrator-3 in Hub-2 to Concentrator-4 in
>Hub-1?

Yes, just as it would have been if the concentrators were standalone.

	paul

    Many thanks for the help, I feel much happier now
    Just to clarify my poorly worded question/example configuration and
    confirm that  no  FDDI bridges are required between the hubs.

    The following shows the two hubs forming the backbone FDDI ring where
    the backplane forms the ring between all of the concentrators in an
    individual hub and the four concentrators shown provide the connections
    between the two hubs completing the ring. Just to check is this OK?

         +-----------------+               +-----------------+  
         | Hub 1           |               | Hub 2           |
         | +---+   +---+   |               | +---+  +---+    |
         | | A |   |   |   |               | |   |  | B |    |
         | | +----------------------------------------+ |    |
         | | +----------------------------------------+ |    |
         | |   |   | B |   |               | | A |  |   |    |
         | |   |   | +-------------------------+ |  |   |    |
         | |   |   | +-------------------------+ |  |   |    |
         | |   |   |   |   |               | |   |  |   |    |
         | +---+   +---+   |               | +---+  +---+    |
         |                 |               |                 |
         +-----------------+               +-----------------+
                                                         
    Sorry if I am asking frequently posted questions but:-
    With the above configuration I take it that I can use the other 5 ports
    on each concentrator for connecting the DAS Dual-Homed stations with
    one connection to each hub from each station (M ports as I understand
    it)?

    Many thanks again
     Neil

   Neil
 
   If this is for the project I think it is, then the network spec/requirement 
   has certainly been upped a little from what we discussed a short while ago 
   eh ?!
 
   I too was a little confused by entry .0 and thought you were coming up with 
   multiple cross-linking of the CONs. However, your diagram -1 clarifes it 
   very well. Your config now even looks just like a plain ol' ring of 
   Concentrators - only you use the hubs to link the colocated CONs instead of 
   fibre cables and release a port on each CON in the process.
 
   The relationships of A-B, B-A is consistent with FDDI connectivity specs.
 
   The end systems can now connect in Dual-Home across separate CONs in 
   separate chassis and provide the resilience you're looking for.
 
   Sorry but I can't resist asking... What if the FDDI adapter in an end 
   system/server breaks ?  How about a second DAS in Dual-Home mode just in 
   case. And then......   Am I being Paranoid or what ?!  ;^)
 
   Are the DECnis' being Dual-Homed too ?
 
   If you're in at UCG on Monday I'll be around if you want to explore any 
   more...
 
   ttfn
 
   Rog