networks, partial mesh, and hierarchical mesh. We will discuss how to design a scalable, robust
internetwork using hub and spoke technology, where and when to use subinterfaces, and so
forth. The first part of this discussion concerns X.25 and how it works. The second part of the
chapter focuses on why NBMA versus subinterfaces is an issue and looks at some of the most
common designs using Frame Relay.
the beach--there are a lot of them. Each carrier might apply different rules to how it moves
traffic throughout the network, and you pay for those services. We don't mention any carriers
specifically, but we mention some of the differences or things to look for with carriers.
the world. It was designed for use over unreliable analog circuits. X.25 is called a reliable
service protocol. It has a reliable data-link layer that uses LAPB (Link Access Procedure
Balanced), and it also provides reliability at the network layer, with the X.25 layer using what
is called PLP (Packet-Level Protocol). Because of this, error checking is used at two different
layers--Layer 2 and Layer 3--which provides for low throughput. LAPB is also flow control-
oriented, which is explained in a moment.
is that some legacy protocols send large packets of more than the 1500 bytes, such as 3000-byte
packets. If these packets cross an Ethernet segment, they must be divided into smaller 1500-
byte frames. As they arrive at their destination X.25 device, they must be reassembled back into
a 3000-byte packet--hence the name Packet Assembler Device.
the way IP is routed. In other words, a source and destination address are built into the packet,
as shown in Figure 10-2. You also can encapsulate IP, IPX, and AppleTalk packets into an X.25
packet, a method commonly called tunneling. And, to make matters even more confusing, X.25
can be encapsulated into IP packets, a tunneling method referred to as XOT (X.25 over
TCP/IP). This topic is discussed in this section.