Create structured addressing plans for IPv4

on

 

Designing a basic subnet plan – IPv4

What is structured adressing plan ?

A stuctured addressing plan is a logical plan on how to design IP subnets for enterprise use.

Why do we need a structured IP plan ?

To provide enough address  capacity to address all nodes in the network depending on the present needs and the future growth plans. Also to summarize the prefixes in the correct locations to avoid gigantic routing tables internally.

What types of devices need IP addresses :

  • Data Networks
  • Wireless LAN’s
  • IP Telephony Networks
  • Video / CCTV Networks
  • Access control systems ( badge readers for example)
  • Network management
  • Server farms
  • Router / switch loopback addresses

One thing a network designer should think of is to assign specific subnets that allow to easily segment communications between different traffic types , for example 192.168.x.x for data and 172.16.x.x for VOIP , it is easier to recognize and to classify, create firewall rules , access-lists etc.

What shoud you consider ?
  • How many devices need to be connected in this network ? Capacity
  • Security ?
    • BYOD ?
    • Corporate ?
  • Stability ?
    • Reduce the broadcast domain as much as you can
  • Gateway IP :
    • Usually the first or the last usable IP of the network – .1 or .254
  • Represent something with each octet
    • 10.0.0.0 =
      • 10 = 10 Class A network
      • Second octet = 0 Site , area, Geographical site
      • Third octet : VLAN
      • Fourth octet : IP address
  • Scalability
    • Depending on the type of the company they may have plans to grow very quickly, then you may need to choose a lower subnet mask to have more availible hosts for the future just in case , but always keeping in mind the stability consideration, don’t double or triple the subnet if it’s not needed because with that comes a bigger broadcast domain and potential issues on that side and also less subnets for other use.

The most important part is to extensively know the list of goals and requirements to make choices based on the clients needs, then to always keep them in mind and make sure regularly that the design meets those requirements

The type of questions a Network Designer should ask before starting a Design draft :
  • What address range is already existing  ?
  • How many locations ? What is the nature of these locations
  • How many devices ?
  • What services are going to be run over the network ?
  • How will addresses be assigned ?
  • Will branches be tied to another site ?

Designing for an enterprise-IPv4

When designing subnets for an enterprise, we are usually working with the three layer campus design model :

  • Access : where all the endpoints and  servers are connected
  • Distribution : aggregation switches of all access switches
  • Core : interconnection of all distribution switches

In this regard we most probably have multiple access zones, geographical areas in out network, what we usually do is choose one big block of ip addresses that we split and divide using VLSM (variable lenght subnet masks) .

example : We have site A with 4 access switches, there are 5 subnets in site A that need a different number of hosts each, we choose a block of addresses , /24 for example and then we chop that down depending on our needs :

  • 10.1.0.0 /24 : Site A
  • 10.2.0.0 /24 : Site B
  • 10.3.0.0 /24 : Site C

As stated in previous post we can see that the second octet is linked to an area, this is for better comprehension and more understandable subnet designs.

After assigning a block of ip addresses for each area, we usually assign VLAN’s on the access layer and use subnets part of that block, as I stated site A needed 5 subnets

  • Subnet 1 : needs 126 hosts : 10.1.1.0/25 VLAN 1
  • Subnet 2 : needs 58 hosts : 10.1.2.0/26 VLAN 2
  • Subnet 3 : needs 2 hosts : 10.1.3.252/30 VLAN 3
  • Subnet 4 : needs 250 hosts : 10.1.4.0/24 VLAN 4
  • Subnet 5 : needs 28 hosts : 10.1.5.0/27 VLAN 5

What we did above is called VLSM, it is the process of dividing a network into subnets of various sizes to prevent wasting IP addresses.

Wait a minute, in the previous post we talked about IP address classes with fixed subnet masks , number of hosts and networks, why are we changing up everything here ? Well with the rapid growth of the internet and the size of the IP routing tables on internet routers we needed a way to improve address space utilization and routing scalability, this is what’s called CIDR (classless interdomain routing).

CIDR moves away from traditional IP classes and defines networks by prefixes and subnet masks

For example : with  IP Class C we had the address space from 192.0.0.0 – 223.255.255.255  which gave us 2,097,152 network blocks with 256 IP addresses each, what if we needed 120 hosts in a network ? In tradtional classful addressing we don’t have a choice each network we choose will have 256 hosts in it, a bit to much don’t you think ?

With CIDR we could say that instead of 192.168.1.0  /24 (255.255.255.0) we could borrow a bit (check subnetting section on previous post) on the right and split the network in half : 192.168.1.0 /25 – 255.255.255.128 which would give us 2 /25 networks with 126 usable addresses each. Exactly what we wan’t !

Back to our practical case, as you can see the third octet is linked to the VLAN number and the subnet mask is manipulated depending on how many hosts we need.

All this is good but what if we use Interior gateway protocols like OSPF, IS-IS, EIGRP ?

That many hosts will mean that there will be a lot of advertisements , the routing tables would be gigantic and if any network goes down, even the smallest one the network will go crazy and need to reconverge, if we advertise each network individually this is not a scalable solution, that’s why we use summarization.

Summarization is taking all the subnets that are used , finding the biggest network they all fit in and only advertise that one, in this case it’s easy as we already did the work the other way, we would advertise 10.1.0.0/24 on the distribution layer and all 5 subnets would be advertised reachable through that rout. Like this if a network goes down it is not advertised to other sites but the convergence will only be done inside the specific site. Other sites will recieve the summary route.

That’s about all you need to know for creating structured addressing plans for IPv4. Thank you for reading and please leave a comment if you have any constructive comments or if this benefitted you in any way !

Comments

Post a Comment