Core/Distribution Submodules

Just finished Chapter 4 of the Self-Study guide.

I understand the basic logic of having a separate distribution/core layer, in fact, thought it made pretty good sense the first time I learned about it in CCNA.

But now that I am seeing all these example diagrams, I really don't see the point of the core layer anymore. From a practicality perspective, not a theory perspective. Especially if we are talking a one-building network or if the dist and core switches are the same L3 stuff, it seems like the core layer is just an unnecessary and extra step. Why not just connect the server farm directly to the dist L3 switches with Gig or 10Gig links?

What is the point of having this fast-switching-focused core layer when there is nothing to switch to, except right back to the dist switch?

Find more posts tagged with

Comments

mikej412

Deadmaster200 wrote:

What is the point of having this fast-switching-focused core layer when there is nothing to switch to, except right back to the dist switch?

There are cases where your core and distribution switches will be the same. When the model meets real life in small locations you will find collapsed backbones or combined core/distribution layers.

Deadmaster200

Yeah, coming from a WAN background, I just really can't see any point of it unless we are connecting multiple buildings or something really really large in one huge building.

By the way, about the topology they give as the answer to the case study question at the end of Chapter 4, is it really necessary and would a company actually request (pay for) such a network when total users over all sites is around only 200 people?

Again, seems no point to actually have a physical core and L3 switches in distribution for such a small number of users. But if I am wrong, I would love a sermon!

mikej412

Deadmaster200 wrote:

is it really necessary and would a company actually request (pay for) such a network when total users over all sites is around only 200 people?

Depends on the customer.

I know a "small" law firm in down town Chicago that has always been tech savvy and has enough hardware to become an ISP or VoIP provider if the law thing doesn't work out for them.

I've seen other companies wait until they hit a growth spurt and then their network (and possibly their business) collapses under the load. Their plan was to grow, but they didn't plan for the growth.

But in real life, small and even medium sized networks usually don't require much design and most Customers aren't willing to pay for it.

mgeorge

This model as described in the CCDA applies to campus wide networks or MAN's.

I used to work at the University of Kentucky and they were just upgrading their core to 10GigE and distrobution uses 1000X and access 100Tx. Although I have to admit during my time at working their, they are slow as hell at doing ****. Which most enterprise campuses are.

It took you over a week possibly two just to get a port activated in a small building of 100 users Which is redicilious. Also one of my students tried to get a communications position there that required CCNA and yet had nothing to do with the CCNA certification. Basically their level 1 network technician just runs cables and activated ports here and there, but mostly they did **** PBX/BELL telephone systems and ocassionaly VoIP and replace 5 foot patch cables.

So yeah personally, working with enterprise networks are not as rewarding as most people want them to be.

Personally I see an enterprise netwok limited by the administration. If the administration are power hungry hogs that want to do everything them selves, then you will notice that it takes forever to get **** done. But if the administration is like me and likes to delegate control then you will notice a fast growing network changing every day to the request of every staff member on campus.

The cisco design track really talks about getting input from personel but out of the 1000's who have their CCDA/CCDP rarely follow the theory and like to make things more complicated then what they really are by just trying to assume what the network needs instead of asking for staff input, besides its not the administrations network, its the companies network and the company is ran by the people so yeah you get my drift.

But any who, as it was said previously, most small busineses have their distro/core mixed. Possibly not even having a distro/core. (thats if the business is small enough) just have 1 router and 1 switch (cisco of course) but if those businesses were to grow, they would hit a serious wall.

Rearden

Here, we have all of the edge switches coming back to a c6513. Each port on the front of our 'core' is to a stack of edge switches. The c6513 deals with all of the intervlan routing and a client's default gateway is really just in the c6513. We're a smallish college (~5k students). I know some colleges that have a 6509 in every building. That seems rather excessive. Is our way the way that things or normally done on campuses or are we rather nonstandard?