Dial Backup

Having a second connection that can move data in the event the primary connection fails is a wise move for today's corporate environment. You can back up a permanent connection or use dial backup, as it's usually called, in several ways to both alleviate congested conditions and provide an alternate route in the event of a link failure.

Dial backup should not be confused with load sharing. Load sharing uses multiple paths to the same destination to send packets. Although dial backup can do this, this feature is usually secondary to making sure that data has a means to get to its destination. Because of this need, dial backup is usually performed on much slower links than load sharing is.

Dial backup can use several different types of interfaces. It's normally configured for an ISDN Basic Rate Interface (BRI) or an analog modem attached to an asynchronous interface, but in advanced setups it might be configured to use a dialer pool or another serial interface.

In addition to backing up a primary link in the event it goes down, a backup interface can also be configured to support a primary line in the event of congestion. The administrator can establish that the backup interface begins to transport data when a given bandwidth threshold on the primary is reached or exceeded.

Configuring Dial Backup for Primary Link Failure

To configure a dial backup link to take over data transport in the event a primary link goes down, you have to do several things. See Listing 11.1 for an example.

Listing 11.1 Dial backup configuration example

 Router(config)# interface serial 0/0 Router(config-if)# backup interface bri0/0 Router(config-if)# backup delay ?        <0-4294967294> Seconds        never          Never activate the backup line Router(config-if)# backup delay 20 ?        <0-4294967294> Seconds        never          Never activate the backup line Router(config-if)# backup delay 20 20 

The tasks can be broken down as follows:

• Identify the primary link, the link currently carrying traffic that you want to back up.

• Identify the dialup link that will back up the primary.

• Configure the primary link (if you haven't already done so) to support the necessary encapsulation, routed protocols, and routing protocols.

• Place the backup interface interface-type number command in the interface configuration for the primary interface.

• Configure how long the backup interface should delay coming up when the primary fails and how long it should wait before dropping when the primary comes back up by using the command:

   backup delay {enable-delay | never} {disable-delay | never} 

The backup interface command options are as follows:

• interface-type number Which interface will back up the primary interface? Remember that interface values vary depending on the router you're configuring. Some routers use the format interface serial 1, whereas a modular router uses the format interface serial 1/1, specifying both the slot and port numbers.

• enable-delay How many seconds must pass before the backup interface takes over for the primary? This setting is a good way to make sure that the primary won't return immediately.

• disable-delay How many seconds must pass before the backup link goes down, once the primary returns? This setting is a good way to make sure that the primary won't drop off again.

• never This keyword prevents the backup line from being enabled or disabled. It is not recommended that you use it for both enable and disable in the same command.

The enable-delay and disable-delay values are the number of seconds you want the interface to wait before coming up or going down. These values can be useful in the event your primary line has a tendency to bounce or go down for just a few seconds occasionally. Although the disable-delay never option has its uses, you'll use the enable-delay never option very few times in a production environment.

Once you configure both the primary and backup interfaces, you place the backup interface in a standby mode. The standby mode simulates a down interface until needed. No traffic passes through the backup interface, and traffic doesn't use the properly configured routes through the interface.

The router with the backup interface configured monitors the status of keepalives, small packets that networking devices send to each other. If carrier isn't detected on the primary link or if keepalives don't arrive from the neighbor router on a regular basis, then your router assumes the primary link has failed and activates the backup.

Floating Static Routes

A static route is a route that someone has manually configured. It tells the router that to reach a certain network you must send data out this interface. Here are two examples of static routes:

 Router(config)#ip route 192.168.4.0 255.255.255.0 BRI0/0 Router(config)#ip route 192.168.5.0 255.255.255.0 172.16.4.1 

The first route points all data that wants to go to the 192.168.4.0/24 network out local interface BRI0/0. Whatever this interface is connected to will be able to forward the data to the destination network. The second instance tells the router to forward all data going to the 192.168.5.0/24 network to a specific remote host, 172.16.4.1. This host might be a device the router is directly connected to or it might be a few hops away. In any case, the router must have a route to this device to forward traffic to it.

Because a router must have a way to determine the best route, it prefers certain routing protocols to others. Cisco routers believe the information provided by static routes is extremely accurate and prefer those routes instead of routes learned via dynamic routing protocols. Table 11.1 shows the relative importance of some routing processes.

Table 11.1. Some Popular Routing Protocols and Their ADs

Routing Protocol	AD
Directly connected	0
Static route reflecting a local interface	0
Static route reflecting a remote device	1
Enhanced Interior Gateway Routing Protocol (EIGRP)	90
Interior Gateway Routing Protocol (IGRP)	100
Open Shortest Path First (OSPF)	110
Routing Information Protocol (RIP)	120

This preference leads to a problem. If you're running a dynamic routing protocol across the primary link on your router, you need to establish a static route to the other networks via the dial backup interface. But if you do so, then the router will prefer to use the dial backup interface rather than the (probably faster) primary interface. A solution exists in the form of a floating static route. In a floating static route, like the one shown in the next line of code, the AD for a static route is changed from 0 or 1 to something higher than the AD for the dynamic routing protocol you're using across the primary:

 Router(config)#ip route 192.168.4.0 255.255.255.0 172.16.4.1 130 

This example shows an alternate route to the 192.168.4.0/24 network. The primary method of getting to the destination network is through BRI0/0, but this command has added a less desirable but still valid option. The router compares the AD of the two routes and use the one pointing to BRI0/0 first because it has the lower AD; 0 is lower than 130. If the BRI interface is ever inactive, the router begins looking for alternatives and uses this route if it has the best AD of those available.

If you have a router running OSPF, the AD is 110. If you set up a static route pointing to an IP network that OSPF already knows about, then the router will prefer to use the path specified by the static route. You can fool the router by changing the AD the static route uses to something higher than the value your routing protocol uses. If you establish a static route and tell the router that the AD for this route is 130, then the router will prefer to use OSPF routes if they are available. This technique allows you to run a dynamic routing protocol across the primary link but have an alternate route via the backup link. The command follows:

 ip route <destination-network> <destination-network mask>   {local-interface | remote device address} <AD> 

The ip route command's options follow:

• destination-network The remote network that this static route refers to. To get to the network, we send packets to the location specified by the rest of the command.

• destination-network mask Generally an IP subnet mask. Helps the router choose between subsets of the same classful IP network number.

• local-interface The local interface by which data needs to leave the router if you want the data sent to this network. This option has an AD of 0.

• remote device address Much the same as local-interface. Specifies the remote device address you need to send data to if you want the data sent to this network. This option has an AD of 1.

• AD The AD for the route you're creating. The AD should be higher than that of your regular routing protocol. The value 130 is good because it's higher than the AD of the four primary routing protocols and, thus, will be used only when the dynamic routing protocol doesn't have an active route.

An example is ip route 10.1.2.0 255.255.255.0 BRI0 130. This command tells the router that to send routes to network 10.1.2.0, it should use interface BRI0. By adding the AD of 130, you make the router compare this route to its existing routing table. In the event of multiple paths to the 10.1.2.0 network, the router uses the active route with the lowest AD. The command ip route 10.1.2.0 255.255.255.0 192.168.1.4 130 also makes the router check its routing table, but instead of sending packets out of the BRI0 interface, the router forwards packets to the device 192.168.1.4. To keep things simple, this device should be a neighboring device.

Activating Dial Backup to Support a Primary Link

You can configure dial backup to activate an interface when the amount of traffic on the primary link reaches or exceeds a certain threshold. Once you select and configure the primary link interface, you need to add this command:

 backup load {enable-threshold | never} {disable-threshold | never} 

Listing 11.2 shows the console output of a router as this command is being entered.

Listing 11.2 Configuring the backup load

 Router(config)# interface serial 0/0 Router(config-if)# backup load ?        <0-100>        Percentage        never          Never activate the backup line Router(config-if)# backup load 70        <0-100>        Percentage        never          Never activate the backup line Router(config-if)# backup load 70 40 

The backup load command's threshold options follow:

• enable-threshold Percentage of use on the primary link at which the backup link will be enabled and start load sharing traffic

• disable-threshold Percentage of use on the primary link at which the backup link will be disabled

• never Prevents the backup link from being enabled or disabled

Unlike many Cisco interface references, the backup load command doesn't use a portion of 255 in the command. The value referenced is a straight percentage of the link's capability.

Do not use values of 0 or 100 in this command. The activation and deactivation values are based on a floating five-minute average, and setting either value to 0 results in a backup link that either always stays up or never comes up.