Ben Piper - CCNP Enterprise Certification Study Guide - Implementing and Operating Cisco Enterprise Network Core Technologies

Figure 1.4 Early Ethernet over a shared medium compared to Ethernet using a switch

Another interesting side effect we inherited from the use of the legendary thick yellow cable is that Ethernet had to impose a limit on the maximum frame size to keep a single node from hogging the medium with colossal frames. The Ethernet maximum transmission unit (MTU) defines the maximum size of the Data field in bytes. DIX and IEEE 802.3 support a maximum MTU of 1,500 bytes. Higher-layer protocols trying to send packets larger than the MTU must break apart their packets into fragments that will fit into the frame's Data field. To avoid fragmentation, some interfaces support jumbo frames with an interface MTU of 9,000 bytes to 9,216 bytes.

When you think of the term subnet, you probably think of an IP subnet address and mask, such as 192.168.1.0/24. The IP subnet address and mask collectively form a CIDR block, or just CIDR for short. But a subnetwork (subnet) is actually a collection of connected nodes that all use the same Data Link layer protocol. For example, a collection of nodes in the same VLAN is an example of a subnetwork. To avoid confusion, I'll refer to the combination of IP subnet address and mask as either a CIDR or an IP subnet.

The moral of the convoluted story behind Ethernet and bridges is that no matter how many tricks and kluges you invent, you can't extend a subnet beyond a few hundred nodes. Regardless of the protocols used, the number of nodes in a subnet is limited by the underlying physical media. To create large networks that include thousands or millions of nodes, we need to join multiple subnets together. This brings us to the Network layer.

Recalling that a subnet consists of connected nodes running the same Data Link layer protocol, the Network layer's primary function is to enable data transfer between nodes that may or may not be in the same subnet. Hence, Network layer protocols must ensure that two things happen:

Nodes in different subnets will communicate using a gateway/router.

Nodes in the same subnet will communicate with one another using the Data Link layer protocol.

It may seem redundant for the Network layer to enable connectivity between nodes in the same subnet, since the Data Link layer already provides this functionality. But the purpose of the Network layer is to abstract the physical and data link characteristics of the network away so that applications don't need to be concerned with them. Instead, the application just deals with Network layer addresses—usually IP addresses.

To see how IP abstracts away the Data Link layer, compare the layer 2 and layer 3 topologies shown in Figure 1.5.

Figure 1.5 Simple layer 2 and layer 3 topologies

IP creates an addressing scheme on top of the Data Link layer, giving each subnet a different CIDR—a combination of an IP subnet address and subnet mask:

VLAN 700—192.168.70.0/26

PPP—10.255.70.0/31

VLAN 706—192.168.70.64/26

A CIDR is the name that IP uses to address a subnet. Hence, a CIDR and subnet should always be tightly coupled, but they're not the same thing. The purpose of a CIDR (IP subnet address and mask) is to help a node determine based on the destination's IP address whether it's in the same subnet or a different subnet. If the destination's IP is in the same CIDR, the node assumes it's in the same subnet and will address the frame to the node's MAC address. Otherwise, the node will assume the destination is in another subnet and will address the frame to the MAC address of the default gateway for the subnet.

The OSI's dream of turning the network into a software abstraction begins to show cracks in the Network layer. Applications do indeed need to have some knowledge of the network, even if it's just IP addresses.

If the destination is in the same subnet, the node will simply communicate with the destination at the Data Link layer. For example, if Host A (192.168.70.3) attempts to ping Router A (192.168.70.62), the following will happen:

1 Host A will note that Router A's IP is in the same subnet.

2 Host A will send an Address Resolution Protocol (ARP) request to the broadcast destination address, asking who has 192.168.70.62.

3 Switch A will flood the ARP request and Router A will receive it.

4 Router A will send an ARP reply to Host A's MAC address. The reply will contain Router A's IP address (192.168.70.62) and MAC address.

5 Switch A will forward the ARP reply to Host A.

6 Host A will encapsulate the IP packet inside an Ethernet frame addressed to Router A's MAC address. Host A will set the Type field in the frame to 0x0800, indicating that the frame contains an IP packet.

A ping is an Internet Control Message Protocol (ICMP) echo request. ICMP is an integral part of IP.

On the other hand, if Host A (192.168.70.3) attempts to ping Host B (192.168.70.67), the following happens:

1 Host A compares its IP address with Router B's IP and determines that they are in different subnets.

2 Host A consults its IP routing table for a closest-match route to 10.255.70.1. Not finding an exact match, the closest match is the default route (0.0.0.0/0). Host A's default gateway is 192.168.70.62, the IP belonging to Router A's Ethernet interface.

3 Because Router A's and Host A's IP addresses are in the same subnet, Host A sends an ARP request asking for Router A's MAC address.

4 Switch A floods the ARP request to Router A.

5 Router A sends an ARP reply to Host A's MAC address. The reply contains Router A's IP address and MAC address.

6 Host A encapsulates the IP packet inside an Ethernet frame addressed to Router A's MAC address. The Type field will contain the value 0x0800 to indicate that the Data field contains an IP packet.

7 Router A receives the Ethernet frame and, based on the Type field in the Ethernet frame, knows it contains an IP packet.

8 Router A looks at the destination IP address in the IP packet and checks its forwarding information base (FIB) for an exact match. Cisco Express Forwarding (CEF) uses the FIB to make forwarding decisions. The FIB is fed by the IP routing table (also known as the Routing Information Base, or RIB). Not finding an exact match for the destination IP address in the FIB, it will use the default route, which has Router B (10.255.70.1) as its next hop.

9 Router A will encapsulate the IP packet in a PPP frame and send it to Router B.

10 Router B will decapsulate the IP packet, look at the destination IP address, and check its FIB for a match. Because the destination IP (192.168.70.67) is in the same subnet as Router B's Ethernet interface, Router B will send an ARP request.

11 Switch B will flood the ARP request to Host B, which will send an ARP reply to Router B's MAC address. The ARP reply will contain Host B's MAC address and IP address.