Few Network Design Questions

[NetworkingStudent1990]

I have posted this on another forum, but I didn't really get the answers I was looking for so I'm posting it here also.

For my final year project I will be designing and implementing a network for a University with over 1000 IP-enabled devices.

I have decided that I would like to implement a routed access layer;  this is something which I haven't done before so I'm not too sure about a  few things, namely:

1. Subnetting. How should this be done? For example, if I have 800  student computers spread out between 17 access switches (800 student  computers / 48 ports per switch) should I simply assign each access  switch "block" a /26 subnet?

2. VLANs. Should each access switch "block" have its own VLAN  assignment? Meaning, for 800 student computers over 17 access switches  there will be 17 different VLANs... if so, how in the real world would  the VLANs be named in order to ensure ease of manageability? I'm  thinking the most logical naming convention would be to simply name the  VLAN after the switch, followed by group, for example, SW1-STUDENT,  SW2-STUDENT and so on.

3. STP, VTP, Trunking... am I right in saying that there is no real use for these technologies in a routed design?

4. FHRP. I want to make use of a FHRP, namely VRRP. I guess the only  place I can implement a FHRP now is for hosts or servers connecting to  multiple L3 access switches for redundancy purposes?

5. For a network this size is there a need to use a three layer model  and make use of blocks? I was thinking of simply having 2 distribution  switches connecting to each of the 24 or so access switches via Gigabit  Ethernet. That would give each access switch (consisting of 48 hosts) 2  Gigabit uplinks. Is this design valid? Is there any reason why 2  distribution switches couldn't handle 1152 hosts (CPU usage, address  tables etc.)?

6. OSPF. Is there a need to divide the network into separate areas? I'm  not sure what the limit is in regards to OSPF neighbours per area.

7. What would be the purpose of interconnecting the 2 distribution  switches? Bearing in mind all access switches are connecting to each of  the distribution switches, and will load balance between them.

Many thanks.

Kind regards,

NetworkingStudent1990

[Leo Laohoo]

1. Subnetting. How should this be done? For example, if I have 800 student computers spread out between 17 access switches (800 student computers / 48 ports per switch) should I simply assign each access switch "block" a /26 subnet?

This is a favorite trick question by the course.  This will determine if the student is being creative or not. 

Define "1000 IP-enabled device"?  Are we talking about PC/Printers?  How about VoIP phones?  What about IP cameras?  And did you consider Wireless access points?

Whatever your answer is, you have a series of choices and this will boil down to the amount of funding.

The simplest method is, yes, put everything in one subnet.  It's not ideal but it would work ... on paper.

In reality, it's no longer recommended because your broadcast domain is HUGE!  Break up the subnet very finely.  Route to every floor. Route every building campus.  For example:

1.  You create each subnet for Staff, student and voice.

2.  You implement routing for every floor of every building in the campus.

[NetworkingStudent1990]

This is not a trick question, as I define the project requirements myself. I am asking HOW should subnetting be done in a L3 access layer design. Common sense tells me that each group of devices connecting to each access switch need to be in their own subnet

Thanks.

[Muhammad Thanveer]

Hi,

I would suggest you to break the lans as

Communication segment (this is for all your switches for example)

If you are considering the Wireless device (put in the vlan in which you are trying to put the wireless users)

Now devide the VLANs for each dept seperatly.(Ex: CSC, EEE,Library,etc...,)

Trunking is must for vlan tagging.

You better go for a static route rather than ospf for this small network

These are only my suggessions rest is up to you..

Think and implement.

Please rate if the info is helpful...

Regards
Thanveer
"Everybody is genius. But if you judge a fish by its ability to climb a tree, it will live its whole life believing that it is a stupid."

[prompt2k2]

Think the following:

1. How many PCs, Printers, VOIP Phones, Cameras

2. Are they sharing any resources (Printers, Servers etc)

3. like the above questions, do you have restrictions on the type of technology involved in your project (Wired or Wireless)

BAsed on the numbers of nodes involved, I think you should use subnetting, group according to faculty or department and not switches, since switches will probably be placed and named according to departments or switches.

4. Will the devices access the internet?

If you can answer these questions, then you might probably have some of your design answers.

[NetworkingStudent1990]

What do you mean by subnet according to faculty/departments? If I were to do that I would have an entire /22 network just for the students... surely that is impossible if you are using a ROUTED access layer because you cannot span the VLANs across multiple switches. Unless your access layer switch connected to a bunch of other L2 access switches, which I guess would kind of defeat the purpose of having a routed access layer in the first place.

Thanks.

[NetworkingStudent1990]

1. How can you span a VLAN over  multiple acces switches if you are using a ROUTED access layer design?  From my understanding, this is not possible.

2. I thought a routed access layer eliminates the need for VLAN tagging.

3.  Why is it better to go with static routes rather than OSPF? I need  redundancy and reliability in this network. The whole point of my thread  was to ask questions regarding how a ROUTED access layer design  works... using static routes would defeat the whole purpose of having a  routed access layer.

Thanks.

[shillings]

You could also check out Layer-3 EtherChannels, so load balancing across multiple uplinks and using them as a single /30 routed link. Think of it as an alternative to IP CEF load balancing or an IGP load balancing. This does require MEC in the distribution or core/distribution layer though (i.e. StackWise, VSS)

Answering one of your previous queries, one way in which static routing is beneficial is that it's a bit simpler to load balance across multiple uplinks using IP CEF. You just need a couple of static default routes at the access layer with same AD. Whereas, an IGP requires injecting the same static default routes, so a little more complex.

[Jeff Van Houten]

Other posters have brought up some valid points, but since this is an exercise I'll throw in my 2 cents given the limitations of the original post.

1. Yes.
2. No. If you are going to route at the access switch, and if you're not going to segregate devices for security or priority, then you might as well use vlan 1 on each switch.
3. Yes.
4. Could be, but an easier choice is probably stacking switches where multiple connections from are given host are needed.
5. Depends on the switch, but in general I don't think a 3 tier model would be needed.
6. No, but since this is a Cisco forum, Cisco would recommend eigrp over Ospf. Much lower convergence times.
7. You would need to interconnect the two distribution switches so that both would have a full view of the routing table.

Sent from Cisco Technical Support iPad App

[NetworkingStudent1990]

Not sure about no. 7. If both distribution switches are connected to the exact same access switches, then surey a link between the distribution switches would never be used... which I guess is a good thing otherwise you'd need to have a high-speed link between the distribution switches as all traffic would traverse through it. Is this incorrect?

Thanks.

[shillings]

I'm thinking that if all layers are routed links and you have cross links between distribution and core layers, then you're right, and a link between distribution switches is of no benefit.

However, if there is a L2/L3 split within the distribution switches, then you would want the link between distribution switches in order to run HSRP.

Does that sound reasonable?

[NetworkingStudent1990]

That's what I was thinking too, just wanted to check though... maybe there is some benefit which I am not aware of.