Load Balancing in Objective-C

Maker PDF417 in Objective-C Load Balancing

EIGRP load balances extremely similarly to IGRP, with the addition of DUAL In EIGRP, if a route meets the FC, it will be entered into the topology table Once all routes to a given destination are in the topology table, the route with the best metric becomes the successor, and the FD is based on its metric Any routes that also meet the metric of the successor times the variance are also entered as successors to the destination, and unequal-cost load balancing is performed across all successors All other routes meeting the FC but not within the variance range are marked as feasible successors To bring this concept home, examine the network shown in Figure 25-45

Figure 25-45: A simple load-balancing example In this network, Queensryche initially hears advertisements for the 10000 network from ONegative, Overkill, and STP The advertised metrics for each of these are shown in the figure, and the local distance is computed as shown here:

Through ONegative 4,500 Through Overkill 6,000 Through STP 3,000

Assuming the variance on Queensryche was configured to 2, Queensryche chooses STP as its successor and sets the FD to 3,000 Because the advertised metric from both Overkill and ONegative meet the FC, they are added to the table ONegative's metric is also less than the variance, so ONegative is entered as a successor as well, and unequal-cost load balancing is performed between both successors Overkill is marked as a feasible successor After building his routing table, Queensryche sends out an update to all neighbors (including Tesla) with Queensryche's new best local metric (3,000) Tesla, in this case, routes to the 10000 network to Queensryche with an advertised metric of 3,500 (assuming, of course, that split horizon is disabled) Queensryche realizes that the path through Tesla does not meet the FC (3,500 > 3,000), and does not add Tesla to the topology table

In contrast to the underlying complexity of DUAL, EIGRP configuration is actually fairly simple (In some ways, it is even simpler than IGRP because there are fewer ways to "blow it up") This section walks through EIGRP configuration and troubleshooting tasks, including

the commands for your reference, and then runs through a few EIGRP case studies to show you how to avoid EIGRP problems This section is broken down as follows:

EIGRP configuration tasks o Enabling EIGRP and using network statements o Changing metric weights o Setting hello and holdtime intervals o Configuring the percentage of link bandwidth that EIGRP uses o Adding a metric offset o Configuring manual summarization o Configuring EIGRP stub routers o Configuring passive interfaces o Configuring EIGRP authentication Redistributing EIGRP o Redistributing EIGRP with RIP, and vice versa o Redistributing EIGRP with IGRP, and vice versa o Redistributing EIGRP into EIGRP EIGRP monitoring and troubleshooting commands EIGRP case studies o Basic EIGRP configuration o Configuring stub routers o Configuring manual summarization o Configuring load balancing

This section briefly describes each common configuration task and the commands that are used in the task You will notice that most of these are very similar to IGRP commands, so little explanation is needed Unlike in other chapters, I do not provide a lengthy example for each command Rather, I use the case studies at the end of the chapter to consolidate configuration and troubleshooting commands into a complete picture of EIGRP operation Enabling EIGRP and Using Network Statements Enabling EIGRP is very similar to enabling IGRP All you need to do is enter the router eigrp [as number] command on each router in the AS, and then use network [network address] statements to enable routing updates to be sent and received and the network(s) in question to be advertised Finally, you may use the default-network statement to specify that a route be used as the default route (just like IGRP) Changing Metric Weights Except that the math is slightly different, metric weights in EIGRP are almost identical to IGRP Like IGRP, by default, K1 and K3 are set to 1, and all other weights are set to 0 (disabled) To change the weights in EIGRP, use the same metric weights [tos] [k1] [k2] [k3] [k4] [k5] command described in 24 One point to remember when changing EIGRP weights, however, is that in EIGRP, routers must be using the same metric weights (and, of course, be configured with the same AS number) to establish adjacencies Setting Hello and Holdtime Intervals

Remembering from the "Operation" section earlier in the chapter, in EIGRP, the hello interval is how often the router sends hello messages (used to establish and maintain adjacencies) to other routers The holdtime interval is how long the router waits to hear a hello from a neighbor before declaring the neighbor dead (causing paths from that neighbor to be removed from the topology table) Although EIGRP does not insist that these intervals be the same on all routers, configuring identical intervals is generally a good idea because misconfigured holdtimes and hello intervals can cause neighbors to be reset at regular intervals (causing routing failures and high CPU use) To set the hello interval on an EIGRP router, use the ip hello-interval eigrp [as number] [interval in seconds] interface config mode command The following example sets a hello interval on the FastEthernet 0/0 interface for AS 10 to be three seconds: