[Search for users] [Overall Top Noters] [List of all Conferences] [Download this site]

Conference 7.286::fddi

Title:FDDI - The Next Generation
Moderator:NETCAD::STEFANI
Created:Thu Apr 27 1989
Last Modified:Thu Jun 05 1997
Last Successful Update:Fri Jun 06 1997
Number of topics:2259
Total number of notes:8590

577.0. "Interlocked tree configuration" by VIVIAN::MILTON (Invisible person it seems!) Wed May 20 1992 07:22

Would the following concentrator configuration work? Supported? Looks to me like
it should under all single failure conditions but I could be wrong! 

Thanks,
Tony.

      +--------+
      |        |a         
      |      +-+-+        
      |      |   +-----------------------------+
      |      | C |m                            |
      |      |   +-------------+               |
      |      +-+-+             |b              |a   
      |        |b            +-+-+           +-+-+    
      |        |             |   |           |   |    
      |        |            m| C |          m| C |    
      |        |          +--+   |     +-----+   |    
      |        |          |  +-+-+     |     +-+-+    
      |        |          |    |a      |       |b     
      |        |          |    +-------+       +----+ 
      |        |          |                         |
      |        |          +----+       +-------+    |
      |        |               |b      |       |a   |
      |        |             +-+-+     |     +-+-+  |
      |        |             |   +-----+     |   +--+
      |        |            m| C |m         m| C |m  
      |        |          +--+   |     +-----+   |   
      |        |          |  +-+-+     |     +-+-+   
      |        |          |    |a      |       |b    
      |        |          |    +-------+       +----+
      |        |          |                         |
      |        |          +----+       +-------+    |
      |        |               |b      |       |a   |
      |        |             +-+-+     |     +-+-+  |
      |        |             |   +-----+     |   +--+
      |        |             | C |m          | C |m  
      |        |             |   |           |   |   
      |        |a            +-+-+           +-+-+   
      |      +-+-+             |a              |b    
      |      |   +-------------+               |     
      |      | C |m                            |
      |      |   +-----------------------------+
      |      +-+-+        
      |        |b         
      +--------+          
T.RTitleUserPersonal
Name
DateLines
577.1DON'T DO THAT!KONING::KONINGPaul Koning, A-13683Wed May 20 1992 10:1923
I can't find anything wrong with it, as far as the rules go.

But that doesn't mean you should do this.

1. This configuration does not (I think) generalize to more than 2 concentrators
   attached to each parent concentrator.

2. Any maps drawn by FDDI mapping programs will look nothing like what you've
   drawn.

3. This configuration bears no resemblance to any normal configuration and
   will be a support nightmare.

4. If you make very small changes to this configuration (swapping some A and
   B connections around) you end up with a VERY different topology, and you
   can easily fragment the network into multiple pieces.  Conversely, in the 
   convential dual-homed tree this IS possible too but is easily avoided.

Don't use that configuration.
It has NO advantages over normal configuration and MAJOR disadvantages.
Use the normal dual homed configuration instead.

	paul
577.2I don't want to!VIVIAN::MILTONInvisible person it seems!Wed May 20 1992 11:2222
Paul,
Thanks for the swift reply, all I can say is I agree with every word but ...

The problem that this customer has is that they are installing fibre plant in 2
adjacent buildings one with 5 floors the other with 7, they want the ability to 
put 2 concentrators on each floor at some time in the future and want to get the 
correct number of fibre pairs. Now this would be simple if they were installing
at least 1 concentrator on each floor but they are not - we can expect runs of 5
floors (with patch panels and patch cables) without concentrators - this worries
me in that I feel that the ring is too exposed. An added complication is that 
they will want to connect to a third building using a British Telecom FDDI 
service where BT control the ring and only supply M-ports (they use a DEC-
concentrator 500s) to the customer. Given this the customer will have to give up
their ring and use SAS internal to their buildings - this increases dramatically
the number of fibre pairs they need installed (they will be using more fibre for
ethernet distribution in the same plant) and so they are looking for a 
compromise.
 
Any help welcome.

Thanks,
Tony.
577.3KONING::KONINGPaul Koning, A-13683Wed May 20 1992 15:1222
As you said, with BT you don't get to see a trunk ring in the first place.
Nor is there (normally) any reason to "expose" the trunk ring anyway.

Pick some place as the "hub"; put a pair of concentrators there, wired into
a tiny trunk ring via their A/B ports.  Wire everything else, dual homed
star fashion, from that point.

Given BT and M ports only at that point, either get two separate BT attachment
points (thus letting BT supply the pair of top level concentrators) or
alternatively just build a dual homed tree attached to a single BT concentrator.

If initially you have concentrators on only a few floors but want to have
stations on every floor, you'll have to run a lot of fiber between floors.
That's why those devices are called "concentrators"!  While fiber is cheap,
terminating lots of fiber isn't, nor is it convenient to deal with lots of
patch panels and other passive (thus not remotely manageable) stuff.
You may be able to use those arguments to justify more concentrators initially.

We can dig deeper on this, but I suspect that may require looking at some
more detailed drawings.  Let me know if you want to do that.

	paul
577.4VIVIAN::MILTONInvisible person it seems!Thu May 21 1992 05:4825
>Pick some place as the "hub"; put a pair of concentrators there, wired into
>a tiny trunk ring via their A/B ports.  Wire everything else, dual homed
>star fashion, from that point.

This is exactly what I suggested to the customer in the first place but given the
number of floors needing concentrators the amount of fibre starts to get out of
hand. 

I have made a decwrite document available in vivian::sys$public:sgw_fddi_1.doc 
that gives a flavour of the problem. 

I am going to see the customer tomorrow and will push for more initial 
concentrators - that may be difficult as they want to start their installation 
with just 4 NSC6800 DAS (NSC have promised them translation bridging by July), no
concentrators and have the ring wander through patch panels on each floor - but 
have enough fibre in place to cover the maximum configuration including the use
of BT service. The comments you have made should help here, but this is a tough 
one. I suppose I could suggest that they install the fibre and terminate it when
needed rather than install patch panels for equipment that may never get 
installed.

Thanks,

Tony.