[c-nsp] Catalyst port groups with trunk ports in between

Eric Kagan ekagan at axsne.com
Thu Mar 30 14:09:48 EST 2006


> A port-channel can either be a VLAN access or a trunk.  It 
> could carry a single VLAN or multiple VLAN's to be the 
> redistributed to other trunks or access ports on the customer 
> switch.  Once the virtual interface has been created by 
> adding ports to the trunk the switch and the programming 
> treats it just like any other interface.

In this case they are Cat 2924/3524XL which use a port-group command on the
interface.  I don't think they create a virtual interface on these switches

Thanks
Eric


> 
> I think that's what you were asking.
> 
> Ryan Dorman
> Millersville University
> 
> On 3/30/06 1:08 PM, "Eric Kagan" <ekagan at axsne.com> wrote:
> 
> >>> From a design perspective, the bonding of two ethernet 
> ports is done 
> >>> from
> >> your equipment to the customer equipment on both sides, something 
> >> like this.
> >> 
> >> 
> >>      Port A ---- Port A                         Port A --- Port A
> >> Cust Switch      Your Switch - Gig Trunk - Your Switch     
> Cust Switch
> >>      Port B ---- Port B                         Port B --- Port B
> >> 
> >> In the above, you would set the Port A/Port A and Port 
> B/Port B up as 
> >> your channel groups using the appropriate configuration 
> (trunk or no 
> >> trunk) for the configuration.  As long as the configuration of the 
> >> ports matches on both sides you should be good to go.
> > 
> > Can the port group be part of a separate VLAN from the switch ?  In 
> > most cases I read, the port group is used for making a 200mb bonded 
> > connection between 2 switches (assuming all other devices 
> on the same 
> > switch are the end points).  In my case, the port group is 
> a separate 
> > VLAN just for a handoff.  None of the other devices are in our 
> > switches.
> > 
> >> 
> >> I'm assuming you want to run multiple GigE links across a 
> single set 
> >> of fibers.  If you can get more fiber then you can just 
> build another 
> >> Port Channel group across the two GigE connections using standard 
> >> GBICS. If you need to run across one set of fibers you 
> will have to 
> >> go with another solution such as CWDM, DWDM, 1310/1550 
> single-fiber 
> >> transceivers, etc.
> > 
> > No, we have the 2 existing fibers terminated in the switch as the 
> > trunk port.  We are just carving out separate 100mb Eth 
> VLANs for different
> > purposes.   In this case we want to bond 2 of the 100mb ports.
> > 
> > Does this make sense ?
> > 
> > Thanks
> > Eric
> > 
> >> 
> >> Hope that helps!
> >> 
> >> Mike
> >> 
> >> 
> >> On 3/28/06 4:57 PM, "Eric Kagan" <ekagan at axsne.com> wrote:
> >> 
> >>> We have the following setup, sample config below.
> >>>  
> >>> We have 2 Data Centers connected via Dark Fiber terminated
> >> on Catalyst
> >>> switches with GBIC's - Gig port is a trunk.
> >>>  
> >>> We have a customer with equipment in both locations so we
> >> hand him a
> >>> 100mb ethernet on his own VLAN.  It been working fine for a year.
> >>>  
> >>> The customer is pushing the 100mb port and wants 200mb (or
> >> GB).  Since
> >>> these are Catalyst 3500XL L2 devices I cannot rate limit 
> or police 
> >>> traffic, so I don't want to give them the 2nd GB port on 
> the switch 
> >>> for fear they will suck up all the bandwidth (they are doing data 
> >>> replication of terabytes of data so I could see this
> >> happening).  We
> >>> do not want to upgrade the switches right now so my idea was add 
> >>> another 100mb port and bond them together.
> >>>  
> >>> So.......we tried a bunch of different configurations and 
> all hell 
> >>> broke loose.  They configured port groups on both their 
> switches at 
> >>> each site.  I tried adding the 2nd FE to the same VLAN 5, 
> no good. 
> >>> Diff VLAN 6, no good. I created port groups on both of our
> >> switches to
> >>> see if we could do a local-local bonding, no good. (We would get 
> >>> either no connectivity across, VLAN mismatches, Relearn
> >> Addr errors,
> >>> etc).  Are we breaking this by being in the middle with 
> VLAN's vs. 
> >>> regular ethernet cables ?  Can this work ? Has anyone ever
> >> done this ?
> >>>  
> >>> If we want to hand then 200mb, what do we need to do on 
> the 2 fiber 
> >>> trunk switches ?  If I can prove out our side, then they
> >> need to prove
> >>> out theirs. (The unknown is whether the customer equipment
> >> would even
> >>> work back to back directly regardless of us in the middle)
> >>>  
> >>>  
> >>> Do all 4 switches need a port group (Side A cust - Side A
> >> us - fiber
> >>> trunk - Side B us - Side B Cust) where we talk to them at 
> each site 
> >>> with a port group ?  Do they still need their own VLAN ?
> >>>  
> >>> or should we be able to just add the 2nd FE to same VLAN 5
> >> ? or hand
> >>> 2nd FE on seperate VLAN (i.e. 6) so it looks like 2 diff Ethernet 
> >>> feeds ?
> >>> 
> >>> Thanks
> >>> Eric
> >>>  
> >>>  
> >>> (Current Config - both sides the same)
> >>>  
> >>> interface FastEthernet0/5
> >>>  description connected to Cust A
> >>>  switchport access vlan 5
> >>> 
> >>> interface GigabitEthernet0/1
> >>>  description connected to fiber-trunk
> >>>  switchport trunk allowed vlan 1,2,4,5,6,7,8,12,1002-1005
> >> switchport
> >>> mode trunk
> >>> 
> >>>  
> >>>  
> >>> (Additional port ??)    --- ??? same vlan ? diff vlan ?  
> port group
> >>> Cust-Colo Loc A  ???
> >>> interface FastEthernet0/6
> >>>  description connected to Cust A
> >>>  switchport access vlan 5
> >>>  port group 5 _______________________________________________
> >>> 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/
> >> 
> > 
> > 
> > _______________________________________________
> > 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/
> 
> _______________________________________________
> 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/
> 




More information about the cisco-nsp mailing list