EIGRP Offset
If I apply an inbound offset value to an interface, shouldn't that only apply to the local metric that is calculated? The AD from the neighbor should not be affected. If apply an offset of 100000 on SW1 fa0/3 inbound, it changes both the FD and the AD.
Before offset:
Rack1SW1#show ip eigrp top 150.1.3.0 255.255.255.0
IP-EIGRP (AS 100): Topology entry for 150.1.3.0/24
State is Passive, Query origin flag is 1, 1 Successor(s), FD is 130560
Routing Descriptor Blocks:
155.1.37.3 (FastEthernet0/3), from 155.1.37.3, Send flag is 0x0
Composite metric is (130560/128000), Route is Internal
Vector metric:
Minimum bandwidth is 100000 Kbit
Total delay is 5100 microseconds
Reliability is 255/255
Load is 1/255
Minimum MTU is 1504
Hop count is 1
155.1.67.6 (Vlan67), from 155.1.67.6, Send flag is 0x0
Composite metric is (130816/130560), Route is Internal
Vector metric:
Minimum bandwidth is 1544 Kbit
Total delay is 5110 microseconds
Reliability is 255/255
Load is 1/255
Minimum MTU is 1500
Hop count is 3
router eigrp 100
offset-list 1 in 100000 FastEthernet0/3
access-list 1 permit 150.1.3.0
after offset:
ack1SW1#show ip eigrp top 150.1.3.0 255.255.255.0
IP-EIGRP (AS 100): Topology entry for 150.1.3.0/24
State is Passive, Query origin flag is 1, 1 Successor(s), FD is 130816
Routing Descriptor Blocks:
155.1.67.6 (Vlan67), from 155.1.67.6, Send flag is 0x0
Composite metric is (130816/130560), Route is Internal
Vector metric:
Minimum bandwidth is 1544 Kbit
Total delay is 5110 microseconds
Reliability is 255/255
Load is 1/255
Minimum MTU is 1500
Hop count is 3
155.1.37.3 (FastEthernet0/3), from 155.1.37.3, Send flag is 0x0
Composite metric is (230560/228000), Route is Internal
Vector metric:
Minimum bandwidth is 100000 Kbit
Total delay is 9006 microseconds
Reliability is 255/255
Load is 1/255
Minimum MTU is 1504
Hop count is 1
If I do the same test on R5 applied to the frame relay multi-point interface connecting to other neighbors, the offset only applies to the local FD and not the AD. Am I missing something?
Comments
Offsets are only supposed to apply to the FD and not the AD. EIGRP offsets work by adding to the delay of the interface when the metric calculation is performed.
That's not true, actually it changes the delay as received inbound, which is part of the AD.That's why it effects both FD/AD (because FD's value is dependant on AD).
This works similar to when it's used on RIP. offset-lists with rip increment hop count when received inbound. While with EIGRP it increments inbound delay only. It actually does a little mathmatical trick. So when we say offset 1000, it will increment the AD delay in a way that the ultimate outcome of the DUAL calculation is +1000.
for this reason, I personally don't like offset-lists with EIGRP. It limits the ratio calculation because it would easily break feasibility condition (RD<FD). Best way use "delay" interface command. If not allowed, then use "bandwidth" interface command, if not then use offset-list as a super rare case.
I'm alost sure we wouldn't need to tweak EIGRP K values to acheive required unequal-cost-loadbalancing ratios. Because others are calculated dynamically (load and reliability). So, as far as ratios are concerned (which i think is your reason to post this question) we have delay, bandwidth and offset-list (yuck!).