Network Plan Starting

The first stage involves looking at the network design at the start of the ISP's operation. Figure 5-12 gives an example network—it has four routers, three switches with some hosts connected to them, and some customer leased-line connections. There is also a dialup router. Finally, the network has a link to an upstream ISP. This is a simple network with four small PoPs at initial rollout.

Figure 5-12. Network Plan at Deployment

Upstream ISP

Figure 5-12. Network Plan at Deployment

Upstream ISP

Network Plan For Isp

5 leascd-l'ine customers

□ unnumbered

5 leascd-l'ine customers

□ unnumbered

15 hosts

Also on the figure are the sizes of the subnets allocated to each portion of the network. In detail, these are as follows:

• WAN point-to-point links have been assigned a /30. There are two hosts on a point-to-point link, so the maximum address space required is a /30. Assigning a larger subnet would result in wasted address space. (If there is trouble calculating how many hosts can fit into a subnet, refer to Table 5-1, which has a few examples of subnetting a /24 address block.)

• LANs have been assigned only the address space that they require.

• One loopback interface is assigned per router.

Table 5-1. Subnet Sizes and Host Counts for a /24 Block

Network Mask


Host Count




























The /30 address required for the ISP's upstream link usually is assigned by the upstream ISP and is not required here. It is assumed that IP unnumbered is used to configure the point-to-point links going to a customer site. It also is assumed that one loopback interface is assigned to each of the four routers. Furthermore, it is assumed that each Ethernet switch has been assigned an IP address for administrative access purposes. These assumptions are common practice in most ISPs for these reasons:

• With IP unnumbered, the point-to-point link between the ISP and the customer routers is not assigned an IP address. Non-Cisco equipment will use other conventions but is similarly capable. Using no address means that there is one less network in the ISP's IGP—IGP design always aims to have minimal networks in the interior routing table for efficiency and convergence speed. (Some ISPs have started to use /31s on point-to-point links—this saves a bit more address space but still means that the address has to be carried around in the IGP. See RFC 3021 for more information on 31-bit prefixes.)

• A loopback interface on a router divorces the router's administrative functions from any of its physical interfaces, thereby guaranteeing continued administrative access in case of any physical link failure.

• All LAN equipment is assigned an IP address for administrative access. Although all equipment will have a serial console, often an IP-capable interface is extremely useful as a first line of entry for administration functions (many ISPs reserve console access as a last resort).

At this initial stage, we have five /30s, one /29, one /28, and two /27s, plus a /30 required for the four loopback addresses (loopback interfaces are assigned a /32). You can follow the sums through in detail here:

2. One /28 and one /29 plus the single /29 make two /28s.

3. The two /28s plus the single /28 make one /27 and one /28.

4. One /27 and one /28, together with the two /27s, make one /26, one /27, and one /28.

5. The next usable address boundary from one /26, one /27, and one /28 is a single /25.

In conclusion, this network infrastructure could be numbered out of a single /25 of address space.

Was this article helpful?

0 0


  • DEAN
    What are essential of routing and network planning?
    3 years ago

Post a comment