One must always weigh the pros and the cons of an L2 versus L3 architecture
to the access layer.
And which features one may need.
To do so based solely on the premise that the current staffing does not know
how to deal with spanning-tree or does not want to because of historical or
lesss than pristine implementation techniques is not a fair basis.
Today mostly all access switch models are capable of L3. So it is a matter
of taste and perspective.
There are positives to both, no doubt. Look at what you are trying to
achieve and what you might lose with either model.
On Tue, Aug 4, 2009 at 10:01 AM, S Malik <ccie.09_at_gmail.com> wrote:
> I would agree with Dale. I think, implementation of pure L3 depends on the
> requirement and any existing architecture (if any).
> In an environment with lot of vlans and STP problems, people may tend to
> convert it to L3 design, however, in a manageable network with few Vlans
> and
> proper configurations (as Dale has mentioned like L2 link b/w Core & Access
> switches), it works fine.
>
> If I need to install 20APs in a building on different floors then at least
> I
> keep them in one Vlan (if they are scattered around few switches).
> For a L3 design, if there is no need to span Vlans across multiple
> switches then I think, L3 is a better solution. One environment is that
> every floor will have its own vlan and there is no requirement of extending
> any vlan across multiple switches then L3 looks a better solution.
>
>
>
>
> On Tue, Aug 4, 2009 at 5:39 AM, Dale Shaw <dale.shaw_at_gmail.com> wrote:
>
> > Hi,
> >
> > On Tue, Aug 4, 2009 at 7:25 PM, backbone
> > systems<backbone.systems_at_gmail.com> wrote:
> > >
> > > can anyone shed some light on this topic.....what are the major
> > > disadvantages of moving ur access switches to layer 3.....performace
> > > should not be counted at all.....
> >
> > It doesn't work very well if you have endpoints with statically
> > assigned IP addresses, because moving a device from one switch to
> > another typically means changing its IP address. Therefore, it doesn't
> > really work within the data centre.
> >
> > I saw one customer try to implement layer 3 uplinks from the access
> > layer switches in a data centre -- each row of racks was served by a
> > pair of 'end of row' switches. This meant each row of racks supported
> > a specific set of IP networks. Equipment is always being shuffled
> > around inside data centres, so the whole system quickly became
> > unworkable and they reverted to a more traditional design.
> >
> > The most common design I see these days is layer 3 in the core and out
> > to distribution switches, with layer 2 links to the access switches.
> >
> > cheers,
> > Dale
> >
> >
> > Blogs and organic groups at http://www.ccie.net
> >
> > _______________________________________________________________________
> > Subscription information may be found at:
> > http://www.groupstudy.com/list/CCIELab.html
>
>
> Blogs and organic groups at http://www.ccie.net
>
> _______________________________________________________________________
> Subscription information may be found at:
> http://www.groupstudy.com/list/CCIELab.html
Blogs and organic groups at http://www.ccie.net
Received on Tue Aug 04 2009 - 11:21:45 ART
This archive was generated by hypermail 2.2.0 : Tue Sep 01 2009 - 05:43:56 ART