cancel
Showing results for 
Search instead for 
Did you mean: 
cancel
867
Views
0
Helpful
7
Replies

EIGRP Load balancing issue

madan.kumar
Level 1
Level 1

Hi

We have topology

LP ----- ET1

|

|

ET2

Where LP is the CORE 6500 and the ET1/ET2 connected to it via EIGRP.As per customer the load balancing is not happening between LP and ET1/ET2.He is seeing only the LP to ET1 is utilized all the time.

I had attached the config for your refference

Pls let me know ur suggestions

7 Replies 7

Giuseppe Larosa
Hall of Fame
Hall of Fame

Hello Madan,

with default K values EIGRP metric is proportional to:

256 * sum (Delays on the path in tens of microsec)

+

256 * 10^7/(lowest BW in the path)

However,

in your case you have modified the bandwidth from LP to ET2 :

interface GigabitEthernet5/6

description ### link betw LP & ET###

no switchport

bandwidth 10240

while link from LP to ET1 has default values:

interface GigabitEthernet5/4

description ### link betw LP & ET###

no switchport

ip address 10.212.176.21 255.255.255.252

routing protocol uses outgoing interface parameters so LP prefers the path via ET1.

to get load-balancing:

or remove bandwidth 10240 in int g5/6 or add it to g5/4

Hope to help

Giuseppe

If you have the bandwidth configured this way for a reason, You can also use the eigrp variance command for unequal cost load balancing

http://www.cisco.com/en/US/tech/tk365/technologies_tech_note09186a008009437d.shtml

or use the "variance" command

lejoe.thomas
Level 3
Level 3

By default EIGRP uses bandwidth and delay to calculate it's metric. In the configuration given, the bandwidth on the interface GigabitEthernet5/6 connecting to ET2 is configured to 10240 kilobits.

If the same routes are advertised by ET1 and ET2. LP will always prefer the routes via ET1 because of the better metric.

In your case,configure unequal cost load balancing using the variance command on LP to load balance across ET1 and ET2.

Using the show ip eigrp topology, you'll see the metrics for the routes advertised by ET1 and ET2. For example, if the metric advertised by ET1 is 20 and that by ET2 is 40 for a route. Using variance 2, the minimum metric now becomes (20 x 2)=40. LP will now choose ET2 and ET1 to get to the route. I hope you the get the idea.

Lejoe

Hi

Thank u all for ur inputs will rate each one of ur postings once i implement this.

As per the customer the ET1 is having 10 mb link and the ET2 is having 20 MB link

should i add variance 2/traffic-share balanced under EIGRP i think that will take care of loadbalancing

Please find the ouputs of

sh ip eigrp topology / sh ip route eigrp pls refer and suggest

lp#sh ip eigrp topology

IP-EIGRP Topology Table for AS(1)/ID(10.212.176.21)

Codes: P - Passive, A - Active, U - Update, Q - Query, R - Reply,

r - reply Status, s - sia Status

P 10.212.177.32/27, 1 successors, FD is 2816

via Connected, Vlan12

P 10.212.190.0/24, 1 successors, FD is 28672

via 10.212.176.22 (28672/28416), GigabitEthernet5/4

via 10.212.176.18 (252928/28416), GigabitEthernet5/6

P 10.212.187.0/24, 1 successors, FD is 26368

via 10.212.176.22 (26368/26112), GigabitEthernet5/4

via 10.212.176.18 (250624/26112), GigabitEthernet5/6

P 10.212.182.0/24, 1 successors, FD is 2816

via Connected, Vlan114

P 10.212.183.0/24, 1 successors, FD is 2816

via Connected, Vlan102

P 10.212.180.0/23, 1 successors, FD is 2816

via Connected, Vlan16

P 10.212.178.0/24, 1 successors, FD is 2816

via Connected, Vlan100

P 10.212.179.0/24, 1 successors, FD is 2816

via Connected, Vlan101

P 10.212.176.0/29, 1 successors, FD is 2816

Codes: P - Passive, A - Active, U - Update, Q - Query, R - Reply,

r - reply Status, s - sia Status

via Connected, Vlan1

P 10.212.176.0/22, 1 successors, FD is 2816

via Summary (2816/0), Null0

P 10.212.177.0/27, 1 successors, FD is 2816

via Connected, Vlan11

P 10.212.168.0/22, 1 successors, FD is 26112

via 10.212.176.22 (26112/2816), GigabitEthernet5/4

via 10.212.176.18 (250368/2816), GigabitEthernet5/6

P 10.212.176.20/30, 1 successors, FD is 25856

via Connected, GigabitEthernet5/4

P 10.212.176.16/30, 1 successors, FD is 250112

via Connected, GigabitEthernet5/6

P 10.212.160.0/26, 1 successors, FD is 26112

via 10.212.176.22 (26112/2816), GigabitEthernet5/4

via 10.212.176.18 (250368/2816), GigabitEthernet5/6

P 10.212.177.192/26, 1 successors, FD is 2816

via Connected, Vlan14

P 10.212.11.0/24, 1 successors, FD is 26368

via 10.212.176.22 (26368/26112), GigabitEthernet5/4

via 10.212.176.18 (250624/26112), GigabitEthernet5/6

P 10.212.0.0/22, 1 successors, FD is 26112

via 10.212.176.22 (26112/2816), GigabitEthernet5/4

via 10.212.176.18 (250368/2816), GigabitEthernet5/6

P 10.212.177.128/26, 1 successors, FD is 2816

Codes: P - Passive, A - Active, U - Update, Q - Query, R - Reply,

r - reply Status, s - sia Status

via Connected, Vlan13

P 199.49.11.0/24, 1 successors, FD is 28672

via 10.212.176.22 (28672/28416), GigabitEthernet5/4

via 10.212.176.18 (252928/28416), GigabitEthernet5/6

lp#sh ip route eigrp

D EX 199.49.11.0/24 [170/28672] via 10.212.176.22, 1d03h, GigabitEthernet5/4

10.0.0.0/8 is variably subnetted, 22 subnets, 8 masks

D EX 10.212.190.0/24

[170/28672] via 10.212.176.22, 1d03h, GigabitEthernet5/4

D 10.212.187.0/24

[90/26368] via 10.212.176.22, 1d03h, GigabitEthernet5/4

D 10.212.168.0/22

[90/26112] via 10.212.176.22, 1d03h, GigabitEthernet5/4

D 10.212.160.0/26

[90/26112] via 10.212.176.22, 1d03h, GigabitEthernet5/4

D 10.212.11.0/24 [90/26368] via 10.212.176.22, 1d03h, GigabitEthernet5/4

D 10.212.0.0/22 [90/26112] via 10.212.176.22, 1d03h, GigabitEthernet5/4

Regards

Madan

Madan,

You need to set the variance to be a value of atleast 10. If you check the show IP EIGRP topology below :

P 10.212.168.0/22, 1 successors, FD is 26112

via 10.212.176.22 (26112/2816), GigabitEthernet5/4

via 10.212.176.18 (250368/2816), GigabitEthernet5/6

You FD is 26112 which is the best metric to reach that destination. Now the same destination via Gig5/6 neighbor is reachable with a metric of 250368. If you want to make this path to be used for unequal cost-path load balancing you have to divide 250368 with the FD 26112 which comes around 10 taking into account the other routes in the EIGRP topology.

HTH, Please rate if it does.

-amit singh

From sh ip eigrp topology

Following are the subnets for which you need unequal cost load balancing.

P 10.212.190.0/24, 1 successors, FD is 28672

via 10.212.176.22 (28672/28416), GigabitEthernet5/4

via 10.212.176.18 (252928/28416), GigabitEthernet5/6

P 10.212.187.0/24, 1 successors, FD is 26368

via 10.212.176.22 (26368/26112), GigabitEthernet5/4

via 10.212.176.18 (250624/26112), GigabitEthernet5/6

P 10.212.168.0/22, 1 successors, FD is 26112

via 10.212.176.22 (26112/2816), GigabitEthernet5/4

via 10.212.176.18 (250368/2816), GigabitEthernet5/6

P 10.212.160.0/26, 1 successors, FD is 26112

via 10.212.176.22 (26112/2816), GigabitEthernet5/4

via 10.212.176.18 (250368/2816), GigabitEthernet5/6

P 10.212.11.0/24, 1 successors, FD is 26368

via 10.212.176.22 (26368/26112), GigabitEthernet5/4

via 10.212.176.18 (250624/26112), GigabitEthernet5/6

P 10.212.0.0/22, 1 successors, FD is 26112

via 10.212.176.22 (26112/2816), GigabitEthernet5/4

via 10.212.176.18 (250368/2816), GigabitEthernet5/6

P 199.49.11.0/24, 1 successors, FD is 28672

via 10.212.176.22 (28672/28416), GigabitEthernet5/4

via 10.212.176.18 (252928/28416), GigabitEthernet5/6

********

Taking one subnet as an example

P 10.212.190.0/24, 1 successors, FD is 28672

via 10.212.176.22 (28672/28416), GigabitEthernet5/4

via 10.212.176.18 (252928/28416), GigabitEthernet5/6

We can see the metric advertised through Gig 5/4 is 28672 and that by Gig 5/6 252928

Smallest metric = 28672

To find variance value

252928/28672 = 8.8 approx 9

Hence, variance value is 9. Just do the above calculation for the subnets to obtain the variance value. As Amit, has suggested, it will be approximately 10. Once unequal cost load balancing is configured, you should see two next hops for the above subnets in the routing table.

Lejoe

Getting Started

Find answers to your questions by entering keywords or phrases in the Search bar above. New here? Use these resources to familiarize yourself with the community:

Innovations in Cisco Full Stack Observability - A new webinar from Cisco