HWIC-D-9ESW (was: Re: [c-nsp] spanning-tree problems)

[Aly, Yasser]

Does this module support bridging with other serial interfaces on the
router using ieee as a protocol?



-----Original Message-----
From: cisco-nsp-bounces at puck.nether.net
[mailto:cisco-nsp-bounces at puck.nether.net] On Behalf Of Brett
Frankenberger
Sent: Saturday, December 10, 2005 5:25 PM
To: cisco-nsp at puck.nether.net
Subject: Re: HWIC-D-9ESW (was: Re: [c-nsp] spanning-tree problems)

On Sat, Dec 10, 2005 at 04:41:11PM +1300, Reuben Farrelly wrote:
> 
> What is most odd is the configuration of this card in IOS.  As Suchy
said, it 
> does not support Rapid PVST which means that your all cisco network
can't 100% 
> interoperate except by falling back to non rapid PVST on ports where
non Rapid 
> devices exist.  VLANS must be configured by going into the vlan
database mode 
> and the VLAN config doesn't show up in the config (in vlan.dat instead
which 
> must be backed up separately).  Also, on trunks you must specify vlans

> 1,1002-1005 as being permitted else it complains about the mandatory
but unused 
> vlans being left out.
> Much the same deal with the WIC-4ESW, although that one is admittedly
a bit older.
> 
> What gives with this?  

There's huge problems with router switch modules (in 28xx's or other
platforms) in any complex topology.  You're restricted to vlan #'s
below 1000; you have to carry VLAN 1 and they have nither UDLD nor
spanning-tree loopguard, which means that a unidirectional path will
often lead to a bridge loop -- if you have a loop in the physical
topology, you can't decide to not have the risk of a bridge loop,
because although you can remove otehr VLANs from trunks to exclude the
possibility of a loop, you can't remove VLAN 1; you're forced to use
the old "vlan database" interface which they have abandoned on nearly
all other switching platforms.  And so on.

After multiple production outages caused by the requirement to carry
VLAN 1 everywhere, I've beat them up pretty hard, but with very little
success.  I think implementation of UDLD is being considered, but
that's it.  Which makes things a little better, but not much.

The bottom line is that there appears to be a strong committment within
Cisco to not treat the ESWs as serious switching platforms.  They work
fine for what I assume is their intended use: install a 2801, connect
it to the world via some sort of serial or Layer 3 Ethernet link, and
then connect 9 or fewer local devices into the ESW.  In that
configuration, spanning-tree is irrelevant, having high VLAN numbers to
match what you already have on the other end is irrelevnat, and so on. 
The ESW is fine for that situation.

> More seriously though, one of the big plusses in buying cisco has
always been 
> the consistency and interoperability between platforms especially the
ones 
> running IOS.  This card has in my opinion been a step backwards in
that regard. 
>   It's still handy to have but it's not even as capable or useful as
the lowest 
> end Catalyst 2940 8 port switch (which I might add as a sidepoint, is
a very 
> nice piece of L2 switching gear at the edge).

When the ESW's first came out (well before the x8xx series routers),
the switching functionality was comparable to what what available on
Cisco's other switching platforms.  Then they started enhancing other
platforms, and one by one, they got the ability to configure VLANs in
IOS, bettern spanning-tree features, support for high VLAN numbers, and
so on.  I assumed they'd eventually get to the ESWs, but apparently
not.

> Are there any plans or thoughts to address this in future versions of
IOS?

I've asked for this.  If more people ask ...

     -- Brett
_______________________________________________
cisco-nsp mailing list  cisco-nsp at puck.nether.net
https://puck.nether.net/mailman/listinfo/cisco-nsp
archive at http://puck.nether.net/pipermail/cisco-nsp/