From what I have found on the internet, there is confusion about the role of MTU in EIGRP metric calculation.
I performed an experiment in GNS with few routers. I was not able to find any change in path cost after the change in MTU value. This was obvious since MTU is not included in the formula of calculating the cost in EIGRP.
Recently I found a "draft-savage-eigrp-00" on ietf explaining the EIGRP protocol. Yet this document also presents some sort of confusion.
Under 'EIGRP Metric Calculation' on page 33, it states that
The composite metric is based on bandwidth, delay, load, and reliability. MTU is not an attribute for calculating the composite metric.
Again on page 35 under 'Wide Metric Vectors' it states that
EIGRP uses five 'vector' metrics: minimum throughput, latency, load, reliability, and maximum transmission unit (MTU).
From my own practical observation I was able to deduce that "MTU is not even used as tie breaking when comparing equal cost paths".
This whole adds up to my confusion. I am not able to deduce the meaning of MTU is a vector and an attribute in EIGRP metric calculation.
I accept that if its not an attribute, it has nothing to do with cost estimation. Hence it is not included in the formula for metric.
metric = {K1BW + [(K2BW)/(256-load)]+(K3delay)}{K5/(reliability+K4)}
If it is considered as one the vectors in find the path, the change in MTU of successor must trigger some sort of process (an update of path if it is too low as compared to feasible successor). But my observation says it does not.
Can someone please help me.