IP Address Range Design - Advice wanted?

[Jellyman_4eva]

Hi,

I am looking at converting a flat layer 2 network which spans multiple buildings in a single geographical area into a layer 3 network...

There are also several independent buildings/sites connected by a variety of means including L2 and L3 links..

At the moment it is a bit of a mess! There are various IP ranges in use everywhere...

What I was thinking of using on the main area was the following:

10.1.x.x/16 - Building 1

10.2.x.x/16 - Building 2

This would then give me a large number of subnets to play with to create simple class C VLANs in each building for per floor PC's, Printers, Management etc.. and would allow fast/easy detection of problem devices.. so

10.1.1.x/24 - Building 1 Management VLAN

10.1.2.x/24 - Building 1 Printer VLAN

Question is should I continue this scheme in the other areas which currently have much smaller numbers of devices (Under 40 for example) for consistency etc, or keep it completely different like:

192.168.1.x

The pro of the above is clear separation, however it appears disjointed, but the con is perhaps no wastage... I could of course use VLSM and really subnet down so could use a 10.10.1.x/24 for one building and 10.10.2.x/24 for another I guess... the question here is, although I am potentially wasting a lot of addresses but not going down to a /24, is that a big deal? Do others do this?

Two areas we connect to actually have several buildings, so I guess I would want to perhaps break off an area for the main area, and another area for each these two sites with multiple buildings, and allocate a range to each of them using a /19 like:

10.1.x.x/19 to 10.31.x.x/19 for main area (With 10.1.x.x as building 1, 10.2.x.x as building 2 as mentioned above)

10.32.x.x/19 to 10.63.x.x/19 at another area

Then at all other areas I could summarize the routes to each area neatly as a single routing table entry...

The other question is around point to point links... I guess I could simply break off the one of the end subnets (10.255.x.x) and break it into many /30 subnets?

Do people do things like this? It all seems a tad too simple for real world use, and that concerns me...

[IAN WHITMORE]

I can only speak for myself, but thats exactly what they do in the real world and in many companies.

Regards,

Ian

[Ven Taylor]

In my experience, I've found it best to stick with /24s wherever possible.  Anything smaller is easy to lose/forget/outgrow.  Anything bigger becomes a waste.

Also, I wouldn't mess with the 192.168 stuff.  Try to keep your summaries as clean as possible.  10.0.0.0 provides over 16 million addresses.  You won't need another subnet if you do it right.

Lastly, give yourself a buffer between subnets to leave room for growth.  Everyone always says they won't grow past "x", but they always do.

I would start with 10.0.0.1/24 and leave it for your managment vlans(switch ip's, router loopbacks, etc).  I would then skip ahead to 10.0.5.0/24 for your user vlans.  Carve out a few-to-several /24s (based on company size and projected growth over the next 10 years)

10.0.15.0/24 through 10.0.20.0/24 could be for printers, phones, mobile wireless devices, etc.

10.0.20.0/24 through 10.0.25.0/24 could be for servers.

I'm sure you get it.

I would also separate my buildings by 2nd octet.  Definitely go with your plan there.

If you find a 10.block that is congruent and unused, I'd definitely start there.  You can re-address slowly by using secondary interfaces, then swinging your statics first.  Once all your statics are good, move your IP helper address so all the dhcp hosts will pull new addresses, then retire the old subnet.

Lather, rinse, repeat for each subnet.

Do it right and you'll have ZERO impact to clients and be a hero with a nice, clean summary!

Best of luck!

Ven

[burleyman]

First off don't worry about waste. These addresses don't cost anything so just think troubleshooting and ease of use.

Here is what we do.

10.xxx.xxx.xxx

2nd octet = Office or Building

3rd octet = Type of equipment ( and possible physical location see below)

4th octet = Just the actual device either staticly assigned or DHCP

So Lets say the main building we will call 1 and lets say there are 3 floors in that building.

IP address as follows

10.1.xxx.xxx - so this means when you see this you know where it is located, Building 1

Now lets say we use this for the users PC's

10.1.101.xxx - this would be for users on the 1st floor of Building 1 and in VLAN 101

10.1.102.xxx - this would be for users on the 2nd floor of Building 1 and in VLAN 102

10.1.103.xxx - this would be for users on the 3rd floor of Building 1 and in VLAN 103

Now lets say you have VoIP also you could do this

10.1.201.xxx - this would be for IP Phones on the 1st floor of Building 1 and in VLAN 201

10.1.202.xxx - this would be for IP Phones on the 1st floor of Building 1 and In VLAN 202

10.1.203.xxx - this would be for IP Phones on the 1st floo of Building 1 and in VLAN 203

Now for network gear we do this. The 3rd octet is always 1 and 4th octet designates the floor.

10.1.1.103 - this would be a switch on the 3rd floor in Building 1

Now printers we do not designate by floor and we do this

10.1.5.XXX - this would be a printer in building 1

We have gone this way and it helps greatly in troubleshooting as we know exactly where the issue is and the type of equipment.

Mike