Looking for help understanding Cisco hierarchical model

[matthew.norman]

Hello all,

I have been trying to get my head around the Cisco hierarchical model but am struggling slightly.

I understand the basic concepts of the layers but not where all of your devices fit in (servers, firewalls, WAN connections etc).

As an example, lets say I have one main building with a LAN and the data center. There would then be various other smaller offices dotted around.

For the LAN I would simply have some access switches connected to a pair of core switches which in turn connect to a pair of routers running HSRP. The routers would act as the gateway for the LAN subnets and any servers only relevant to the LAN would connect to the LAN core.

In this setup are my routers considered part of the core layer or the distribution layer or merged in to one?

When it comes to the data center, this is where I get more confused.

Firstly, would you typically have you local WAN connections coming in to the two LAN routers or would you typically have them coming in to the data center.

Then at the data center level we would have the core layer and the distribution layer.

Which layer would I connect all of my servers to?

Which layer would my firewalls sit at with my DMZ and connection to the internet?

If I was using layer 3 switches instead of routers, would I use the core switches or the distribution switches for this?

Apologies if these sound like silly questions I just need a little guidance.

I fully understand that not all networks strictly follow the Cisco hierarchical model however I would prefer to stick with this as the starting point of my design learning.

Appreciate any advice.

Matt

[Leo Laohoo]

I don't know where to start.  

Traditional hierarchy is core, distribution (aka "distro") and access.  

Access layer connects to distro layer and distro connects to core.  But with the advent of Nexus FEX/Instant Access the hierarchy changes dramatically.  The physical core switch turns into a three-in-one, i. e.  core, distro and access, logically and physically speaking.  

Servers, for example, can connect to a Nexus 2K.  A Nexus 2K is designed as a top-of-rack switch and is controlled and managed by a parent switch in form of a Nexus 5K/6K or 7K upstream.  So this means the parent switch becomes core, distro and access layer.  

One of the biggest stumbling block with implementing a hierarchy model is funding and lack of space.  This is why some downstream clients connect directly to the core switch instead of an access layer switch.  

[matthew.norman]

Thanks Leo,

That clears it up a little.

I have been looking at various network designs and a lot of them seem to be as you have described where a Nexus is used as the core with the fabric extenders being used for server access etc.

Regards

Matt

[Joseph W. Doherty]

Disclaimer

The Author of this posting offers the information contained within this posting without consideration and with the reader's understanding that there's no implied or expressed suitability or fitness for any purpose. Information provided is for informational purposes only and should not be construed as rendering professional advice of any kind. Usage of this posting's information is solely at reader's own risk.

Liability Disclaimer

In no event shall Author be liable for any damages wha2tsoever (including, without limitation, damages for loss of use, data or profit) arising out of the use or inability to use the posting's information even if Author has been advised of the possibility of such damage.

Posting

It might help if you first understand the original purpose of the hierarchical model was for scalability.  Once you have need for more than two devices, how do you interconnect them?  A hierarchy is one possible topology, and often the preferred topology.

The 3 layer hierarchy model can also scale rather large, so you seldom will need more than 3 layers.

Functionally, hosts connect to the bottom layer, the "access" layer.  Traditionally that layer usually was only L2, but today L3 might be extended to the access layer.

If you have 3 layers, the top layer is the "core", generally it's devoted to "high" speed L3, and the intermediate layer is "distribution", which might be L2 or L3.  This intermediate layer often tries to insure the core doesn't need to do anything but route traffic.  For example, if might filter traffic with ACLs.  (Also keep in mind, traditionally L3 processing was more "intensive" than L2 processing, so obtaining fast L3 performance was more of a problem.  I.e. the distribution layer tried to preclude the core layer from having to do anything beyond routing.)

If your network size didn't really need all 3 layers, 2 layers topologies were used, often combining the core and distribution layers, known as a collapsed core.

For devices like WAN connections, and other non-host devices, they generally don't connect to access layer devices.  They might connect to distribution or core layer or might even be a direct link to the core.

"Old" data centers were often treated just other edge hosts, but "modern" data centers are often now a collapsed distribution/access layer.

Here's a Cisco design guide that might help you: http://www.cisco.com/c/en/us/td/docs/solutions/Enterprise/Small_Enterprise_Design_Profile/SEDP/chap2.html

When reading documents like the prior, don't get hung up too much on what's being done, but try to understand why it's being done.  When you being to understand the "why", hierarchal designs should make much more sense, and also when to use them, and when you don't need to use them.

A 3 layer hiearchal design is not always the best design.