Regarding the above post i believe that formulas provided by Cisco documentation are not clear in terms of metric calculation.
I spent several time doing tests and understand how to derive the exact metric but i obtained different results.
let's look at it together please.
i did a try first with links above 1GB and then under 1GB assuming at same time to exclude bandwidth out of calculation so the command was:
eigrp metric weights 0 0 0 1 0 0 (you should use 2 different formulas in function of link speed as provided from Cisco DOC but simply didn't worked).
What i noticed is that you can find out the metric using always the same formula which i
found also incomplete and i believe i understood why.
FROM R6 CHECK METRIC FOR R9 VLAN 9 INTERFACE:
R6#show eigrp address-family ipv4 topology 150.1.9.9
255.255.255.255
EIGRP-IPv4 VR(MULTI-AF) Topology Entry for AS(100)/ID(150.1.6.6)
for 150.1.9.9/32
State is Passive, Query origin flag is 1, 1 Successor(s), FD is
1392640, RIB is 10880
Descriptor Blocks:
155.1.67.7 (GigabitEthernet1/0.67), from 155.1.67.7, Send flag
is 0x0
Composite metric is (1392640/737280), route is Internal
Vector metric:
Minimum bandwidth is 8000000 Kbit
Total delay is 21250000 picoseconds
Reliability is 255/255
Load is 1/255
Minimum MTU is 1500
Hop count is 2
Originating router is 150.1.9.9
R6#show ip eigrp topology
155.1.9.0/24
EIGRP-IPv4 VR(MULTI-AF) Topology Entry for AS(100)/ID(150.1.6.6)
for 155.1.9.0/24
State is Passive, Query origin flag is 1, 1 Successor(s), FD is
1966080, RIB is 15360
Descriptor Blocks:
155.1.67.7 (GigabitEthernet1/0.67), from 155.1.67.7, Send flag
is 0x0
Composite metric is (1966080/1310720), route is Internal
Vector metric:
Minimum bandwidth is 10000 Kbit
Total delay is 30000000 picoseconds
Reliability is 255/255
Load is 1/255
Minimum MTU is 1500
Hop count is 2
Originating router is 150.1.9.9
ok now let's look first example: Assuming the link speed is more than 1Gig we find comp metric (10^7*65536/10)/8000000 = 8192
we have 21250000 picoseconds , result do not match.
Look instead the other formula : TOT DELAY * 65536 /10 =
196608000000
now the other : TOT DELAY * 65536/10 = 139264000000
It seems quite close but still far from the result.
The reasons why formulas sound strange i believe are 2: first is that first link
is above 1 GB! we are currently up to 8GB now.and we should use
the first formula(with bandwidth) but i think should be used not
in the way Cisco provided, i still don't know how.Also Cisco provided (107*65536/10)/BW which also do not match.
Second, in my personal opinion here we should divide not for 10 but for picoseconds contained
in 1 microsecond. That is because in wide metrics EIGRP think in
picoseconds and represents delay for calculated metrics in
topology table in picoseconds but the interfaces still work delay
in microseconds. If you take any above metric and divide for 1
million you obtain the correct metric.
1 million of picosecond is 1 microsecond . By doing this operation
instead dividing by 10, every metric now report the correct result.
30000000 * 65536 /1000000 = 1966080
21250000 * 65536 /1000000 = 1392640
Any thought about this consideration would be appreciated, thank you