| T.R | Title | User | Personal
Name | Date | Lines |
|---|
| 1988.1 | | NPSS::TAYLOR | | Tue Mar 19 1996 13:25 | 24 |
|
It's not really upstream-downstream but perhaps more what port
types are being physically connected together. Some data is exchanged
between the neighboring phys when you first make a physical connection.
Among other things the information will include port type (A,B,M,S).
I'm a little confused with your illustration and this is what
I believe your topology looks like.
station_1 port B to concentrator_1 port M
station_1 port A to concentrator_2 port M
station_2 port B to concentrator_1 port M
station_2 port A to concentrator_2 port M
concentrator_1 port B to concentrator_2 port A
Your stations dual home based on the phy types and connection rules.
It's up to the A/B ports to accept or reject the connection. Station_1
and station_2 will understand that there are connections to it's A and
B ports, both the neighboring (concentrator) phys are M ports. Based on
connect rules station_x will accept the B to M connection and reject the
A to M reason being STANDBY.
|
| 1988.2 | | STRWRS::KOCH_P | It never hurts to ask... | Tue Mar 19 1996 15:03 | 11 |
|
This is the configuration with a little more detail. The question is
how does the near end A detect that the far end B is active and go into
standby mode?
DAS DAS DAS DAS
Concentrator1 Station Station Concentrator
M M M M M M <-> A <-> B <-> A <-> B <-> M M M M M M
A B <-------------------------------> A B
^ ^
|---------------------------------------|
|
| 1988.3 | | 35356::RABAHY | dtn 471-5160, outside 1-810-347-5160 | Tue Mar 19 1996 17:08 | 14 |
| re .2:
That is probably not the configuration you really want. Below is probably what
you do want;
+------------------+
+-----------------M| DAS concentrator |M--------------+
A +------------------+ A
+-----------+ B A +-----------+
| station 1 | | | | station 2 |
+-----------+ A B +-----------+
B +------------------+ B
+-----------------M| DAS concentrator |M--------------+
+------------------+
|
| 1988.4 | | STRWRS::KOCH_P | It never hurts to ask... | Tue Mar 19 1996 18:30 | 9 |
|
re: .3
Sorry, but it is. It represents what happens when you have two
DECswitch 900EFs in a DEChub 900. It is also how a 3COM LANplex works.
So, if the stations were physically seperate, yes, .3 could apply.
However, if for example this was 2 VAX-4000s with FDDI controllers, why
would I waste M-ports? .3 essentially consumes 4 M-ports, but .2 only
consumes 2 M-ports.
|
| 1988.5 | topology rules | NETCAD::ROLKE | Tune in, turn on, fail over | Wed Mar 20 1996 10:33 | 50 |
| Redraw the configuration from .2
######################################
# ###################### # <-- Dual Ring
# # # #
+------------+ +-------------+
| A Con1 B | | A Con2 B |
| M M M M M | | M M M M M |
+------------+ +-------------+
| |
| +----------------+ | <-- Trees
| | | |
+-------+ +-------+
| A B | | A B |
| Das1 | | Das2 |
+-------+ +-------+
The FDDI topology matrix defines the connection rules.
If Das1A was connected and Das2B was connected then what would be the
the topology of the Logical ring? Answer: that doesn't work. Das1A knows
he can't connect to Con1M because of the topology rules so he goes to
standby.
Note .3 suggested:
######################################
# ###################### # <-- Dual Ring
# # # #
+------------+ +-------------+
| A Con1 B | | A Con2 B |
| M M M M M | | M M M M M |
+------------+ +-------------+
| | | |
| | +------------------+ | <-- Trees
| | | |
| +-|----------------+ |
| | | |
+-------+ +-------+
| A B | | A B |
| Das1 | | Das2 |
+-------+ +-------+
In this case Das1 and Das2 are "dual homed". In both cases if all connections
are good then both Das stations will connect via their B ports to Con2 and
both A ports to Con1 will be in standby. The M ports are not "wasted", rather
this is the price you pay for dual homing.
Regards,
Chuck
|
| 1988.6 | | STRWRS::KOCH_P | It never hurts to ask... | Wed Mar 20 1996 15:44 | 7 |
| re: .5
"wasted" was a poor choice of a word.
In regard to the top diagram of .5, what is used on the "wire" to tell
the DAS1a not to connect to CON1m? I really that's the rules, but there
must be some sort of "messaging" which occurs to make it happen?
|
| 1988.7 | SMT does NIFfing | NETCAD::ROLKE | Tune in, turn on, fail over | Wed Mar 20 1996 16:28 | 19 |
| This messaging is the job of SMT, FDDI Station Managment.
When you connect any two FDDI ports they go through several layers of
protocol to bring up a connection. (Surely this is written up somewhere?)
After symbol passing and link error testing then the SMT entities on the
two ends of the connection decide if the connection joins the ring or not.
The link has acutally come up far enough to pass SMT frames. One type of
SMT frame is NIF, neighbor information frame, whereby each node can tell
his partner about his capabilities.
So even though a link looks like it is in standby it may well have traffic
on it!
SMT is pretty complicated in that there is a ton of details spread around in
firmware and hardware. However, each aspect of SMT taken by itself is
straightforward. I wish I understood it better.
Chuck
|