Network Products

Updated: 17 April 1997
 

IP Switching Tech Tip


Assigning IP Addresses in IP Switching


Introduction

When planning the installation of an IP Swithcing network, you need to work out the assignment of addresses for the IP Switch Controller, and the associated IP Switch Gateways. Your goal is to take a small chunk of contiguous addresses from your site's IP address space, and carve this chunk up into tiny point to point subnets that connect each Gateway to the Controller in a private subnet.

[Diagram]

The diagram above shows a basic IP Switching installation consisting of three IP Switch Gateways and one IP Switch Controller. The Gateways in this example are routing traffic from three Fast Ethernet subnets, but that is not the focus of this note. We're interested in the addresses assigned to the ATM interfaces that connect the Gateways and Controllers to the GIGAswitch/ATM.

Procedure

  1. Determine the number of Gateways in your installation. For each Gateway you will need FOUR contiguous IP Addresses. The first address in the list of four must be evenly divisible by four.

    Basically, you're only concerned with the last digit of your IP addresses in this case. IP addresses are of the form: blah.blah.blah.NUMBER you care about the last number. In the example above, the TOP gateway got assigned the addresses blah.blah.blah.4, blah.blah.blah.5, blah.blah.blah.6, and blah.blah.blah.7 To keep this simpler, we'll say it got addresses 4,5,6, and 7.

    The key is that the base address 4 is evenly divisible by 4.

  2. Make a table that looks like the following:

    IP AddressBinary EquivalentRole
    *.*.*.400000100Network Number
    *.*.*.500000101Controller
    *.*.*.600000110Gateway 1
    *.*.*.700000111Broadcast Address

    If you examine the diagram above, you'll note that the Controller was assigned address 5 and the Gateway got address 6. This assignment of addresses and their roles is the consequence of using a 30-bit network mask, as instructed in the GIGAswitch/IP Installation and Configuration manual. A 30-bit IP address mask leaves just TWO bits to identify the nodes in a subnet. The binary equivalents shown in the table demonstrate the effect of the netmask. The 30-bit netmask leaves only the last two binary digits (shown in bold type in the table) for the node number. So, node 00 is unused, node 01 becomes the Controller node, node 10 becomes the Gateway node, and 11 is the Broadcast address for this tiny subnet. Since the subnet is point to point, the broadcast address is unused.

  3. Extend your table to cover all of your Gateways:

    IP AddressBinary EquivalentRole
    *.*.*.4 00000100Network Number
    *.*.*.5 00000101Controller
    *.*.*.6 00000110Gateway 1
    *.*.*.7 00000111Broadcast Address
    *.*.*.8 00001000Network Number
    *.*.*.9 00001001Controller
    *.*.*.1000001010Gateway 2
    *.*.*.1100001011Broadcast Address
    *.*.*.1200001100Network Number
    *.*.*.1300001101Controller
    *.*.*.1400001110Gateway 3
    *.*.*.1500001111Broadcast Address

Notes

In the above example, we had three Gateways, and got a block of 12 contiguous addresses. This is the ideal arrangement, but is not absolutely necessary. The only requirement for the assignment of addresses is in step one of the procedure. The least significant digit of the base address of the block of four addresses must be divisible by four.