- A virtual community of network engineers
 Home  BookStore  StudyNotes  Links  Archives  StudyRooms  HelpWanted  Discounts  Login
Re: L3 routed network [7:125087] posted 08/14/2007
[Chronological Index] [Thread Index] [Top] [Date Prev][Date Next] [Thread Prev][Thread Next]

In a pure L3 environment, you do not need to worry about any L2
redundancy protocol.  Your EIGRP configuration will take care of all
the redundancy.  This is the benefit of using a pure L3 deployment.
You will end up with zero loops, so you don't have the issue of STP,
plus all your redundancy/loadbalancing is handled inside your
much-easier-to-troubleshoot EIGRP configuration.


On 8/14/07, Bill Wharton  wrote:
> I just want to point out that the only L3 broadcast domain is a single
> access switch itself and localised at that. The access switch would be
> configured with 'no switchport' and hence connected with a point-to-point
> routed link to the distribution. Same thing for distribution->core and all
> over the network
> Is your response still the same?
> Thank you
> ""Neal R""  wrote in message
> news:200708141450.l7EEoooX019131@xxxxxxxxxxxxxxxxx
> > Each access switch should attach to two distribution switches. One
> > distribution switch is primary for HSRP and the other secondary. This
> > scheme follows through on the back(core) side, by adjusting routing
> > metrics such that the primary HSRP switch is favored in both spanning
> > tree and routing metric calculations. Core/distribution is a little less
> > clear cut and depends greatly on what resource consumer/producers exist
> > and where they are.
> >
> >  The access, distribution, and core layers are drawn with separate
> > switches in the Cisco propaganda. They could also be called the layer in
> > which gross profit comes from many low margin devices, the layer in
> > which gross profit comes from some medium margin devices, and the layer
> > in which high gross profit is achieved with pairs or quartets of very
> > high margin modular devices. Notice that there isn't a "you get a nice,
> > fat raise layer" anywhere in there ...
> >
> >    Access, distribution, and core are *functions*, not separate
> > concentric rings of switches. It is very common to find access and
> > distribution functions collapsed with core functions found in routers
> > which also aggregate WAN links. It is even more common to find lots of
> > 29xx and 35{12|24|48} which need the distribution/core function
> > installed to partition an overlarge broadcast domain. A network needing
> > separate access, distribution, and core must be very large indeed, or
> > have certain geographic characteristics, such as a multibuilding campus
> > in which access and distribution are done per building along with
> > access, distribution, and core functions found in the building housing
> > the datacenter functions for the campus.
> >
> >     Stop teasing and tell us how many widgets are in each physical
> > location and what the interconnection is like :-)
> >
> >
> >
> > Bill Wharton wrote:
> >> L3 routed network and usage of HSRP,GBLP,etc
> >>
> >> I'm designing a L3 routed network - L3 from core -> distribution ->
> > access.
> >> Hence, the only L2 would be the local access switches themselves and
> > VLAN's
> >> would not span a single access switch or 3750 cluster
> >>
> >> Now, would I need to use HSRP/GBLP/VRRP anywhere in the design? I don't
> > see
> >> a reason to anywhere. Some access switches may connect to a single
> >> distribution while other access switches may be dual-homed to two
> >> distribution switches. However, with dual-homing, I'll simply use EIGRP
> >> equal cost paths
> >>
> >> Also, all distribution switches would be dual-homed to the two cores.
> >>
> >> Anybody differ?
> >>
> >> Pg10 of this guide states I should be using one of these redundancy
> >> protocols hence the confusion
> >>
> >

Message Posted at:
FAQ, list archives, and subscription info: