Subnetting question

[agathodaimon]

I'm working on assigment, and here it goes:

"The company wants to use the 10.0.0.0 private IP address.
The network spans 6 buildings, and each building has about 300 workstations each.
Design a simple addressing solution that leaves ample room for growth and is easy to administer. Explain what happens to your design if the number of hosts per network jumps to over 1,024 per building."


Now, how do we treat this "private" ip address? Because when I mask 8 bits, that gives me over 250 networks and over 65,000 hosts.
That seems excessive to me... Or is that the correct way to treat this?

Any help?

 

[reicherb]

We aren't going to do the assignment for you. Share what you are thinking. There are many ways to accomplish this. The 10.0.0.0 address block will allow over 65000 addresses. You have to decide where to break them up. In my organization we assign a class B subnet to each building.

 

[NickOlsen8390]

Not really sure how to answer your question.
I would break it up across the buildings, giving each building like a class b, or 2-3 class c's.
give the servers there own class c also. And put a router at the network entrance to each building, being the gateway address for there ip's.
to break it down.
you could have 10.0.0.0/22 in building 1 10.0.4.0/22 in build 2 and so on. Each buildings router being its respective gateway(10.0.0.1, 10.0.4.1)
And doing no nat on any of the routers. Or even having each building router come back to its own interface on a main router would work good. The main router also having the internet connection.

 

[agathodaimon]

I wasn't expecting you to.

I just thought that since we only need 300 workstations in each building, and there are only 6 buildings, that the number of subnets to hosts is a bit excessive.
Would that be considered normal use? Or am I missing something?

I want to mask off 8 bits for the subnets, then leave 16 bits for the hosts. That is what my mind is coming up with....

 

[reicherb]

We give each building a Class B and then break that up by type of host. We give servers 1 class C, PCs, 2-3 class Cs, video equipment a class C, printers a class C, reserve a class C for other devices that need a static address (boiler controls, access control, etc).

We actually use a pretty small percentage of the class B for each building but logically it seems to make sense.

 

[ScottMac]

Start by looking at the smallest power of two that covers your need (in this case, it's 512 .... "over 300 with ample room")

How many bits to 512? That's your mask per subnet.

You can add multiple subnets to a building, or do some additional subnets from your initial slice.

Remember that you'll also need some minimal subnets for point-to-point links, backup, and administrative space, etc

Other strategies involve using parts of the address to label the location, such that (for example) any errors popping up in the monitors that start with 10.1.X.X come from systems in building #1, 10.2.x.x from building 2, etc.

It's private address space, you can slice it and dice it any way you want.

Good Luck

 

[Qacer]

10.0.0.0
xxxxxxxx.xxxxxxxx.xxxxxxxx.xxxxxxxx
11111111.11111111.1aaa0000.00000000

1's represent the bits from the private address that you want to retain.
a's represent the bits allotted for the sub-network address.
0's represent the bits allotted for the host address.

In that case, I would use the following network addresses:
1). 10.0.128.0/20
2). 10.0.144.0/20
3). 10.0.160.0/20
4). 10.0.176.0/20
5). 10.0.192.0/20
6). 10.0.208.0/20

... And you would still have one more sub-network address to spare. If you use this addressing scheme, then each of your sub-networks will have [2^(12) - 2] = 4094 hosts.

My way of coming up with this addressing scheme is simple. First, I determine how many hosts each of my sub-networks need. I convert this number into binary. Then, I use the number of bits in the binary value to reserve the rightmost bits. In other words, I count starting from right of the binary-converted IP address and see how many "left-over" bits I can use for the sub-networks.

Then, I decided how many sub-networks I need and reserve the necessary bits in the IP address. In other words, I start counting again from the right, but my starting point would be the bit position after the hosts bit count.

I hope this helps.

 

[her209]

10.0.0.0/24 gives you (2^8-2) 254 hosts per network
10.0.0.0/23 gives you (2^9-2) 510 hosts per network
10.0.0.0/22 gives you (2^10-2) 1022 hosts per network
etc.

 

[drebo]

That question is horribly worded. There are so many possible solutions to that question that solve the first part without having to deal with the second part.

Do the subnets need to be contiguous? If not, then expanding them when the network grows is easy. If yes, then they need to be big enough to start with and you're going to waste a lot of addresses.

Is the zero subnet rule in effect? If so, that determines where you can start.

Etc. Etc.

Other than that, it's a fairly academic question. Figure out how many host bits you need to accomodate your 300 hosts and work from there.

 

[agathodaimon]

Ok thanks for all the helpful replies...

I agree that the question is horribly worded... As are a lot of things in this book.

But I'm now beginning to realize how this private IP can be used, thanks to all the examples ^^^. Once I have time to sit down today and finish it, I'll post it here. I'm sure it's much simpler than I made it out to be, but subnetting is something I really want to nail down.

 

[xSauronx]

Originally posted by: agathodaimon
I'm sure it's much simpler than I made it out to be, but subnetting is something I really want to nail down.

good place to get some practice

 

[reicherb]

What did you come up with?

 

[Tarrant64]

Maybe I'm not remembering correctly, but my instructor for my TCP/IP course was looking for something rather simple for a solution.

10.0.0.0 (building 1)
10.1.0.0 (building 2)
10.2.0.0 (building 3)
10.3.0.0 (building 4)

...and so on for what you need. It's easy to manage, and virtually more than enough hosts than you'll ever need. Try not to think about how excessive it is, just think "simple". I may actually have a post here or on another forum with this exact same problem. I'll see if I can dig it up, many others had solutions for this as well.

So 10.0.0.0/16, that notation could be completely wrong though. I need to hit the books again.



I could be wrong though.

HERE:
http://www.hardforum.com/showthread.php?t=1265761

 

[Tarrant64]

Originally posted by: agathodaimon
I'm working on assigment, and here it goes:

"The company wants to use the 10.0.0.0 private IP address.
The network spans 6 buildings, and each building has about 300 workstations each.
Design a simple addressing solution that leaves ample room for growth and is easy to administer. Explain what happens to your design if the number of hosts per network jumps to over 1,024 per building."


Now, how do we treat this "private" ip address? Because when I mask 8 bits, that gives me over 250 networks and over 65,000 hosts.
That seems excessive to me... Or is that the correct way to treat this?

Any help?

The easy to administer part is really what comes into question. An experienced person in networking would probably be able to handle many different subnetting configurations, however I think that the problem you have there is more of a beginner's problem, meaning they don't need you to come up with a complex answer(not that the ones provided above are...just trying to think of the simple school text book answer).

Originally posted by: agathodaimon
Ok thanks for all the helpful replies...

I agree that the question is horribly worded... As are a lot of things in this book.

But I'm now beginning to realize how this private IP can be used, thanks to all the examples ^^^. Once I have time to sit down today and finish it, I'll post it here. I'm sure it's much simpler than I made it out to be, but subnetting is something I really want to nail down.

This is good! If you truly want to pursue a career in networking, you will need to know this. My instructor has put a subnetting question as the final question on most tests I have taken so far. Each one different, sometimes complex or very simple(although not worded to make you realize that).

 

[Tarrant64]

Did you ever get a solution to this?

 

[heymrdj]

Originally posted by: Tarrant64
Maybe I'm not remembering correctly, but my instructor for my TCP/IP course was looking for something rather simple for a solution.

10.0.0.0 (building 1)
10.1.0.0 (building 2)
10.2.0.0 (building 3)
10.3.0.0 (building 4)

...and so on for what you need. It's easy to manage, and virtually more than enough hosts than you'll ever need. Try not to think about how excessive it is, just think "simple". I may actually have a post here or on another forum with this exact same problem. I'll see if I can dig it up, many others had solutions for this as well.

So 10.0.0.0/16, that notation could be completely wrong though. I need to hit the books again.



I could be wrong though.

HERE:
http://www.hardforum.com/showthread.php?t=1265761

Sometimes simple is best. :thumbsup:

 

[spidey07]

Originally posted by: heymrdj
Sometimes simple is best. :thumbsup:

Yeah, but that's a terrible use of address space and very wasteful.

10.0.0.0/22 gives 1022 addresses
10.0.8.0/22 next net
10.0.16/22 next net
10.0.24.0/22 next net
10.0.28.0/22 RESERVED

This still leaves the following nets open
10.0.4.0/22
10.0.12.0/22
10.0.20.0/22
10.0.28.0/22

So if any of the original nets need to be expanded to a /21 you can do so by just changing the mask and accomodate 2046 hosts. So addresses the "what happens if you go over 1024" question and it scales and doesn't waste address space. And can be summarized nicely with 10.0.0.0/19 which would be 10.0.0.0 - 10.0.31.254.

 

[Tarrant64]

Originally posted by: spidey07

Originally posted by: heymrdj
Sometimes simple is best. :thumbsup:

Yeah, but that's a terrible use of address space and very wasteful.

10.0.0.0/22 gives 1022 addresses
10.0.8.0/22 next net
10.0.16/22 next net
10.0.24.0/22 next net
10.0.28.0/22 RESERVED

This still leaves the following nets open
10.0.4.0/22
10.0.12.0/22
10.0.20.0/22
10.0.28.0/22

So if any of the original nets need to be expanded to a /21 you can do so by just changing the mask and accomodate 2046 hosts. So addresses the "what happens if you go over 1024" question and it scales and doesn't waste address space. And can be summarized nicely with 10.0.0.0/19 which would be 10.0.0.0 - 10.0.31.254.

Yeah, it is. But it didn't say you have to be efficient. Just simple.

The problem above is from Chapter 1 or 2 of a TCP/IP if I remember correctly, and they haven't gone into some of the things that are possible with subnetting to make designing a network more efficient. So this, again, is more of a beginners answer.

We all know it would never pass for CCNA.

 

[drebo]

Right, but starting out with bad practice is not a good idea. I recently ran across a new client whose previous IT guy set them up on the local subnet of 100.10.10.0/24.

Sometimes, wrong is wrong for a reason. Best practice should be taught from the very beginning.

 

[her209]

Originally posted by: drebo
Right, but starting out with bad practice is not a good idea. I recently ran across a new client whose previous IT guy set them up on the local subnet of 100.10.10.0/24.

Sometimes, wrong is wrong for a reason. Best practice should be taught from the very beginning.

Is this wrong because its not using a private address space?

 

[drebo]

Yes.

 

[reicherb]

So, what answer did you use?

 

[Tarrant64]

Originally posted by: reicherb
So, what answer did you use?

That's what I still want to know too.

 

[reicherb]

Maybe he never did the assignment.....