Source Route Bridging

IBM developed SRB in the mid-eighties as a way to bridge between Token Ring local-area networks (LANs). In SRB, the source determines the route to arrive at the destination node before sending an information frame to it. This differs from Ethernet transparent bridging, where the bridges build and maintain tables containing paths to data-link layer destinations. In SRB, the source node acquires the routes to destinations on the network using explorer frames.

In Figure 3-1, when a source node wishes to send information to a destination that is not on the local LAN, it sends out an explorer frame. Bridges receive the explorer frame in one interface and forward the frame out all other interfaces. Bridges add route information, called route descriptors, to the frames as they travel throughout the network. The route information includes each bridge and ring number over which the explorer frame has traveled (see Figure 3-2). In Figure 3-2, when the explorer frame reaches Bridge 6, it adds information about Ring 7/Bridge 6 and the outgoing Ring 3. At Bridge 4, route information of Bridge 4 and Ring 2 is added. The bridge value of the last route descriptor is set to 0, to indicate the destination has been reached. The information gathered by the exploration process makes up the routing information field (RIF) of a Token Ring frame.

Host A Ring 7 Bridge 6 Ring 3 Bridge 4 Ring 2 Host B

Ring 7

Bridge 6

Ring 3

Bridge 4

Ring 2

Bridge 0

RIF Routing Information

RIF Routing Information

When the explorer frames arrive at the destination, the destination node sets the direction bit (also known as the D-bit) to 1 and sends the frame back to the source node via the same route it used to arrive at the destination. When multiple reply frames reach the source node, it usually uses the route of the first frame received. Other decision metrics include the minimum number of hops and the path with the largest maximum transmission unit (MTU) allowed.

Unlike transparent bridging, which uses the Spanning-Tree Protocol to form loop-free paths to data-link layer destinations, SRB is inherently loop-free and does not rely on the Spanning-Tree Protocol when delivering data. SRB can use Spanning-Tree Protocol when sending explorer frames, thus reducing the traffic generated during route discovery.

The routing information field is contained in the 802.5 frame. It is composed of a 2-byte RIF header and may contain one or more route descriptors (see Figure 3-3).

Figure 3-3 Routing Information Field

Type

Length t

Unused

D-bit

12-bit Ring Number

4-bit Bridge Number

In the RIF first byte, the first 2 bits indicate the type of frame, as follows:

00—Indicates this is a regular frame that should be routed to the destination following the information in the route descriptors.

10—Indicates this is an all-rings explorer frame that should take all possible routes to the destination.

11—Indicates this is a Spanning-Tree explorer frame that should take only one path to the destination.

The next bit is unused. The 5 least significant bits in the first byte indicate the total length of the RIF field, including the 2-byte header. A RIF with only one bridge hop is 6 bytes long.

The most significant bit in the second byte is the D-bit, which determines whether the route descriptors in the RIF header are processed left-to-right or right-to-left. A value of 0 indicates left-to-right. A value of 1 indicates right-to-left.

As displayed in Figure 3-3, the next 3 bits indicate the largest frame size that can be handled along a designated route (011 indicates 4472 bytes). The lower 4 bits are not used.

As indicated in Figure 3-3, the route descriptors are 2 bytes long. The first 12 bits are used to indicate the ring number. The 4 least significant bits are used to indicate the bridge number.

SRB Design with Cisco Routers

SRB is limited in that it is a flat network topology in the data-link layer. Another limitation is the maximum hop count of seven bridges. Token Ring chipsets were originally designed to process two rings. Routers can be introduced into the network to help overcome these limitations. This is accomplished with the concept of the virtual ring (see Figure 3-4). Consider each router interface as a minibridge that connects the external ring to the internal virtual ring. The ring number is used in the RIF field as in any other physical ring. To configure the virtual ring group, use the global command source-bridge ring-group virtual-ring-group-number. Each Token Ring LAN is bridged to the virtual ring using the interface command source-bridge ring-number bridge-number virtual-ring-group-number.

Figure 3-4 Virtual Ring

Traffic from Ring 1 to Ring 3 flows through Bridge 1, then Ring 5, and finally Bridge 3.

source-bridge ring-group 5

Traffic from Ring 1 to Ring 3 flows through Bridge 1, then Ring 5, and finally Bridge 3.

source-bridge ring-group 5

The virtual ring can be extended to several routers in a cloud, to connect remote Token Rings (see Figure 3-5). Although there are several router hops between Token Ring 1 and Token Ring 2, the RIF field for a frame will only show two bridge hops.

Figure 3-5 Extended Virtual Ring

Figure 3-5 Extended Virtual Ring

0 0

Post a comment