This is actually not the case.  A layer two network that contains a logical loop with 
get into trouble with all types of traffic flows.  For example, consider a simple 
network like the one I'll try and draw below.


Node A                  Node B
   |                        |
---------Ethernet 1---------------------------
        |       |
     Bridge A   Bridge B
        |       |
--------------Ethernet 2----------------------

Consider a transmission from Node A to Node B and assume that both nodes were recently 
powered up and that the bridge tables for the two bridges have not been built.

Here is a likely scenario.  The frame hits Ethernet 1 and both bridge A and B detect 
the packet and check their forwarding tables only to realize they do not have an entry 
for B.  At this time they add Node A to the forwarding table as residing on Ethernet 
1.  They then both forward the frame onto Ethernet 2.  Node B also receives the frame 
and processes it.  However, 2 versions of the frame now exist on or headed two 
Ethernet 2.  When the bridges forward to Ethernet 2, (1 at a time) each bridge will 
process the others frame and decide that A has indeed moved to Ethernet 2 and adjust 
their forwarding tables appropriately.  Assuming B hasn't responded as yet, they will 
also process each others frames and still not finding B in their forwarding table, 
copy them back onto Ethernet 1.  However, when they do, they will again switch A's 
address from Ethernet 2 to 1 and likely repeat the process again.  

Just think what happens when B responds :)

All in all, this is very unhealthy.  The fact that there is no way at layer 2 in 
ethernet to detect the age of frames severely hinders loop mitigation.  In fact, there 
pretty much isn't any loop mitigation at layer 2 in ethernet networks.  If you create 
a looped topology and do not prune the redundant paths, you will bring a network down 
in seconds.

Pete
 

*********** REPLY SEPARATOR  ***********

On 2/18/2001 at 4:58 PM Jack Yu wrote:

>Actually, the main reason to eliminate multiple paths is because of
>broadcast at layer 2. Lay 2 devices have to this forward broadcast, and
>multiple paths to a single destination will cause broadcast storm. Layer 3
>devices do not have this problem only because they do not forward broadcast,
>they either read it or drop it. So if there is a network with no layer 2
>broadcast, you do not need spanning tree at all. Of course, you can also
>disable it when you are sure there is no duplicated path exists.
>
>Regards,
>Jack
>
>
>
>
>""Groupstudy"" <[EMAIL PROTECTED]> wrote in message
>[EMAIL PROTECTED]">news:[EMAIL PROTECTED]...
>> Spanning Tree's job is to eliminate multiple paths to a single
>destination.
>> If it finds more than one path it will put one of them into blocking mode
>to
>> ensure a loop free path.   Remember, Spanning Tree runs at layer two and
>has
>> no concept bandwidth.  If you need to setup equal cost paths to a certain
>> destination, you will need to do it at layers above layer two.
>>
>> Pick up a copy of Radia Perlmans 'Interconnections' second edition.  Radia
>> is the primary authority on the Spanning Tree algorithm, she will
>enlighten
>> you.  This is also good solid study for deeper networking knowledge.
>>
>> ----- Original Message -----
>> From: AndyD <[EMAIL PROTECTED]>
>> Newsgroups: groupstudy.cisco
>> To: <[EMAIL PROTECTED]>
>> Sent: Sunday, February 18, 2001 8:58 AM
>> Subject: Equal cost switching
>>
>>
>> > Spanning tree is supposed to choose the one best switched path.  But if
>> you
>> > set up two equal cost paths, will it use both?  Is there a way to force
>it
>> > to use the bandwidth from both paths?
>> >
>> > Thanks!
>> >
>> >
>> > _________________________________
>> > FAQ, list archives, and subscription info:
>> http://www.groupstudy.com/list/cisco.html
>> > Report misconduct and Nondisclosure violations to [EMAIL PROTECTED]
>> >
>>
>> _________________________________
>> FAQ, list archives, and subscription info:
>http://www.groupstudy.com/list/cisco.html
>> Report misconduct and Nondisclosure violations to [EMAIL PROTECTED]
>>
>
>
>_________________________________
>FAQ, list archives, and subscription info: http://www.groupstudy.com/list/cisco.html
>Report misconduct and Nondisclosure violations to [EMAIL PROTECTED]



_________________________________
FAQ, list archives, and subscription info: http://www.groupstudy.com/list/cisco.html
Report misconduct and Nondisclosure violations to [EMAIL PROTECTED]

Reply via email to