Rate and give feedback:  Feedback Received. Thank You!

Rate and give feedback: 

X

This document helped resolve my issue

Yes No

Additional Comments

800 characters remaining

May we contact you if necessary?

Name:  

E-mail: 

Enter a valid Email ID

Need product assistance? Contact Juniper Support

Submit

This site is protected by hCaptcha and its Privacy Policy and Terms of Service apply.

English  

English

Network Addressing at Layer 2 and Layer 3

Both Ethernet and IP use globally unique network addresses that can be used as the basis for a truly global network. Ethernet MAC addresses come from the IEEE and IP subnet addresses come from various Internet authorities. (IP also employs a naming convention absent in Ethernet, but we'll ignore that in this discussion.) The key differences in how these addresses are assigned make all the difference when it comes to the basic functions of a bridge as opposed to a router.

Note: The opposite of a “globally unique network address” is the “locally significant connection identifier” which connects two endpoints on a network. For example, MPLS labels such as 1000001 can repeat in a network, but a public IP address can appear on the Internet in only one place at a time (otherwise it is an error).

• MAC addresses are 48 bits long. The first 24 bits are assigned by the IEEE and form the organizationally unique identifier (OUI) of the manufacturer or vendor requesting the address. The last 24 bits form the serial number of the LAN interface cards and their uniqueness must be enforced by the company (some companies reuse numbers of bad or returned cards while others do not).
• IPv4 addresses are 32 bits long. A variable number of the beginning bits are assigned by an Internet authority and represent a subnet located somewhere in the world. The remaining bits are assigned locally and, when joined to the network portion of the address, uniquely identify some host on a particular network.
• IPv6 addresses are 128 bits long. Although there are significant differences, for the purposes of this discussion, it is enough to point out that there is also a network and host portion to an IPv6 address.

Note that MAC addresses are mainly organized by manufacturer and IP addresses are organized by network, which is located in a particular place. Therefore, the IP address can easily be used by routers for a packet's overall direction (for example, “192.168.27.48 is west of here”). However, the MAC addresses on a vendor's interface cards can end up anywhere in the world, and often do. Consider a Juniper Networks router as a simple example. Every Ethernet LAN interface on the router that sends or receives packets places them inside Ethernet frames with MAC addresses. All of these interfaces share the initial 24 bits assigned to Juniper Networks. Two might differ only in one digit from one interface to another. Yet the routers containing these MAC interfaces could be located on opposite sides of the world.

An Internet backbone router only needs a table entry for every network (not host) in the world. Most other routers only have a portion of this full table, and a default route for forwarding packets with no entries in their table. In contrast, to perform the same role, a bridge would need one table entry for every LAN interface, on host or bridge, in the world. This is hard enough to do for Ethernets that span a metropolitan area, let alone the entire world.

Note: There are other reasons that Ethernet would be hard-pressed to become a truly global network, including the fact that MAC addresses do not often have names associated with them while IP addresses do (for example, 192.168.27.48 might be host48.accounting.juniper.net). This section addresses only the address issues.