Help designing a Class A network, using 10.0.0.0

T

Tarrant64

Weaksauce
Joined
Mar 1, 2007
Messages
95
I've been trying to wrap my head around designing a simple class A network that is easy to manage and allows extra room for growth. This is part of a case project for homework, and I've attempted to do some of the work but I am lost as to what I may be doing wrong.

Basically, I have 6 buildings in which I need to setup a Class A network for(1 router in each building). I have to use a 10.0.0.0 private IP address. There approx. 1,000 workstations between the 6 buildings, the max in one being 312.

I was trying to design a class A network thinking I only needed 6 subnets. Well, doing the math to find out how many bits I need to borrow, I find that 4-bits gives me enough subnets that I need. That's 2^4-2, giving me 14 usable subnets. Simple, right? Well, I waste a HUGE amount of host addresses that way.

The problem I was running into was I only wanted to configure about 2,000 hosts per site. Reason being a questioned was also asked what happens with my design should the hosts ever reach 1,024 'per' building. This should take care of that. Well, if I lower the hosts per subnet, I get a large amount of subnets, right?

On a 10.0.0.0 network with a 255.240.0.0 subnet mask, I end up with a little over 1 million hosts.


On a 10.0.0.0 network with a 255.255.248.0(trying to keep it at 2,048 hosts per subnet), I end up with a lot of subnets.

Is this something I have to deal with, with it being a class A network? I am trying to find more on designing class A networks, but there isn't much on it. There are several combinations using subnet masks I'm just not figuring out how to do the simple one I want.

Also, is CLSM or VLSM better for this configuration? All the examples I have for CLSM are for Class B and C networks.

Thanks in advance!!!

Please feel free to ask for anymore details.
 
Keep in mind that larger subnets usually means more hosts to generate broadcast traffic, and more MAC addresses for your switches to handle. Most switches I have seen tend to cap out around 8000 unique MAC addresses. I would suggest going with smaller subnets. This will also give you room to grow as the theoretical company grows, with so much waste on smaller locations.

I don't know if you have covered this yet, but you can have variable length subnets, they don't have to all be the same size. How ever as I said you may not have covered that in your class yet.

BTW, I have seen the problems large subnets cause. The University I attend has a /16 allocated to them, and the "user" traffic (excluding wifi) is on a /17. A little excess broadcast traffic will kill that network simply because of how many freaking hosts are actually on it, and the switches (which are Enterasys) have hit their MAC address limits and started having to spit traffic out onto ports it doesn't belong.

EDIT:

Read your reply again, some how I missed the VLSM mention. Use variable length subnets, the IETF came up with it to help deal with the inefficiency of classful routing.
 
Here's a VLSM calculator:
http://www.vlsm-calc.net/

It looks like you're on the right path, just basically use what you need. All the subnet masks do is help you design how to organize and shape your network traffic.
 
That's VLSM calc you posted Keiichi was pretty nifty. I am still going through trying to find mroe on VLSM. I think what I'm looking at right now is what I had before, with a /21 network.

So 10.0.0.1/21. That's 10.0.0.1 with a 255.255.248.0 subnet mask.

My 6 networks would be:

10.0.8.0
10.0.16.0
10.0.24.0
10.0.32.0
10.0.40.0
10.0.48.0

This allows for 2,046 hosts per building, which is more than enough room for growth. does that sound right? It doesn't look like I truly use VLSM this way though.
 
Actually it's more like this.
Your WAN (or main network) that connects all six subnets together will be 10.0.0.0/18 then you can subnet as such:
Subnet A = 10.0.0.0/21
Subnet B = 10.0.8.0/21
Subnet C = 10.0.16.0/21
Subnet D = 10.0.24.0/21
Subnet E = 10.0.32.0/21
Subnet F = 10.0.40.0/21

All VLSM is is not using the default subnet masking.

Edit:
After rereading I realized this was more of a textbook problem. and here's the answer:
Subnet A 10.0.0.0 /22 255.255.252.0 10.0.0.1 - 10.0.3.254
Subnet B 10.0.4.0 /22 255.255.252.0 10.0.4.1 - 10.0.7.254
Subnet C 10.0.8.0 /22 255.255.252.0 10.0.8.1 - 10.0.11.254
Subnet D 10.0.12.0 /22 255.255.252.0 10.0.12.1 - 10.0.15.254
Subnet E 10.0.16.0 /22 255.255.252.0 10.0.16.1 - 10.0.19.254
Subnet F 10.0.20.0 /22 255.255.252.0 10.0.20.1 - 10.0.23.254

With the subnet gateways being the first IP of the subnet for each subnet. This gives the ability to scale to "1024" hosts (yes there is a reason for the quotation marks)
 
Keiichi said:
Actually it's more like this.
Your WAN (or main network) that connects all six subnets together will be 10.0.0.0/18 then you can subnet as such:
Subnet A = 10.0.0.0/21 255.255.248.0 10.0.0.1 - 10.0.7.254
Subnet B = 10.0.8.0/21 255.255.248.0 10.0.8.1 - 10.0.15.254
Subnet C = 10.0.16.0/21....
Subnet D = 10.0.24.0/21...
Subnet E = 10.0.32.0/21...
Subnet F = 10.0.40.0/21....

All VLSM is is not using the default subnet masking.

Edit:
After rereading I realized this was more of a textbook problem. and here's the answer:
Subnet A 10.0.0.0 /22 255.255.252.0 10.0.0.1 - 10.0.3.254
Subnet B 10.0.4.0 /22 255.255.252.0 10.0.4.1 - 10.0.7.254
Subnet C 10.0.8.0 /22 255.255.252.0 10.0.8.1 - 10.0.11.254
Subnet D 10.0.12.0 /22 255.255.252.0 10.0.12.1 - 10.0.15.254
Subnet E 10.0.16.0 /22 255.255.252.0 10.0.16.1 - 10.0.19.254
Subnet F 10.0.20.0 /22 255.255.252.0 10.0.20.1 - 10.0.23.254

With the subnet gateways being the first IP of the subnet for each subnet. This gives the ability to scale to "1024" hosts (yes there is a reason for the quotation marks)
Click to expand...


Thanks again for the response.

With 10.0.0.0/22 network I am only taking into account 1,024 hosts, if I needed to expand beyond that I have the additional subnets I can use, correct? 10.0.24.0/22 and up, right?

As I understand it VLSM basically allows you to configure the network so each subnet only supports the amount of hosts needed, so you're not wasting additional host addresses. This also allows for future expansion with the available subnets left.

Would my original answer(bolded above) still be a solution?
 
just a few notes.

i do not know of anyone who uses classful routing still. everyone pretty much uses CIDR now (VLSM). also, you don't lose the first and last subnets when you are breaking up a network. back in the day, i think routers either couldn't handle it for some reason or it was merely not recommended to use them. however, all current routers can handle it, and the CCNA has finally been updated (i'm pretty sure anyway) to not use the old answers of losing subnets. you still lose two hosts per subnet (network and broadcast), but you don't lose full subnets.

anyway, from what you are saying it sounds like it wants the answer given above. however, in most environments today if you have 6 buildings that each get their own subnet, you would give each one enough space for each host and include some expansion room. if one building has 1000 hosts, while the other 5 only have 50 each, it's stupid to allocate a huge subnet to the other 5 buildings.
 
I would also like to mention, you don't have to break all of your network into equal sized networks, you can have a /17 and a /24, just as long as they don't overlap you're fine. Use this to your advantage and allocate what is necessary and will meet your projected growth.
 
Can anyone recommend a good book on this subject or the most updated techniques?
Thanks!
 
OK guys, I finished up that case project last night and turned it in on time. The solution I had was the 10.0.0.0 /22 network. Seen above you can see what the rest of the layout would have been for the answer.

I also took the chance to talk to the professor about the case project in general as I was also in contact with him back and forth via email about the solution and what exactly he was looking for. He said while that the solutions I had may very well work, it wasn't simple enough. And this is what we ended up coming up with after looking at the alternatives.

Using a 10.0.0.0 /16 (that's 255.255.0.0 subnet mask), which gives 65k+ hosts per network. However, what makes this easy to administer is this.

Network ID 1: 10.1.0.0
Network ID 2: 10.2.0.0
Network ID 3: 10.3.0.0
Network ID 4: ....
...
...


He was looking for each building to be pretty much designed around this(building 1 being network ID 1 and so on). This makes it easy to administer. Reason being that should there be a problem with a host, on say 10.2.4.7, I would know where to look, building 2. We had talked about broadcast storms among other interferences on a network, and that this would make it simple to determine on what network these events occured on. There were other things talked about as well, but I guess I was thinking more complex about the problem.

He said in this point of view the administrator wouldn't be worrying about the "wasted" host addresses that would most likely never be used. I was on the right track as far as making the network more efficient, but the 10.0.0.0 /16 was the easiest solution.

Thanks for everything and the support! I did learn a lot more and in the future should there be a need for a more complicated, robust, however you want to say it...network I will be more prepared. He also told me that my research into VLSM and more subnetting will help out a great deal in the Cisco courses I will take in the future.
 
Tarrant64 said:
OK guys, I finished up that case project last night and turned it in on time. The solution I had was the 10.0.0.0 /22 network. Seen above you can see what the rest of the layout would have been for the answer.

I also took the chance to talk to the professor about the case project in general as I was also in contact with him back and forth via email about the solution and what exactly he was looking for. He said while that the solutions I had may very well work, it wasn't simple enough. And this is what we ended up coming up with after looking at the alternatives.

Using a 10.0.0.0 /16 (that's 255.255.0.0 subnet mask), which gives 65k+ hosts per network. However, what makes this easy to administer is this.

Network ID 1: 10.1.0.0
Network ID 2: 10.2.0.0
Network ID 3: 10.3.0.0
Network ID 4: ....
...
...


He was looking for each building to be pretty much designed around this(building 1 being network ID 1 and so on). This makes it easy to administer. Reason being that should there be a problem with a host, on say 10.2.4.7, I would know where to look, building 2. We had talked about broadcast storms among other interferences on a network, and that this would make it simple to determine on what network these events occured on. There were other things talked about as well, but I guess I was thinking more complex about the problem.

He said in this point of view the administrator wouldn't be worrying about the "wasted" host addresses that would most likely never be used. I was on the right track as far as making the network more efficient, but the 10.0.0.0 /16 was the easiest solution.

Thanks for everything and the support! I did learn a lot more and in the future should there be a need for a more complicated, robust, however you want to say it...network I will be more prepared. He also told me that my research into VLSM and more subnetting will help out a great deal in the Cisco courses I will take in the future.
Click to expand...

yea, he's wrong...

but i could see why he'd want it to be simple for you... i mean you could have totally used a class C, and i probably would have... depending on the projected growth...

it was just a school problem with a school answer
 
goodcooper said:
it was just a school problem with a school answer
Click to expand...
That is dead on. In the real world you HAVE to think about unused space, because if you run out of addresses and go to ARIN asking for more, they will just laugh at you when you tell them your efficiency is so low in spots.
 
Tarrant64 said:
We had talked about broadcast storms among other interferences on a network, and that this would make it simple to determine on what network these events occured on.
Click to expand...

I have to disagree with your prof. here ... the huge subnets will be the cause of the broadcast storms, so very limited in usefulness as a diagnostic tool ... "hey I tracked down the problem to the x.x.x.x /16 subnet, and to fix it we need to segment the network". :face palm:

Anyway, I will point out that this is beyond the scope of the 'simple assignment', but real world, the excuse of diagnosing broadcast storms is bogus. There are lots of ways to monitor and mitigate broadcast storms, and prevention (segmentation aka small subnets) is the first step.

You would be much better off knowing that a problem or issue exists in the accounting department on floor 5 of building 2 on switch x in the headquarters location rather than the problem exists somewhere at headquarters.
 
You must log in or register to reply here.
Back
Top