Outbound Load Balancing

[network_dude]

Hey Everyone,

I have 4 T1s which I am load balancing via ip cef per destination. That seems to be working for for my receiving, but my transfer does not seem to be load balancing. Please see my "sh int sum" below:

nterface IHQ IQD OHQ OQD RXBS RXPS TXBS TXPS TRTL

------------------------------------------------------------------------

* FastEthernet0/0 0 0 0 0 254000 434 5604000 478 0

FastEthernet0/1 0 0 0 0 0 0 0 0 0

* Serial0/1/0:0 0 0 0 0 1409000 125 0 0 0

* Serial0/2/0:0 0 0 0 0 1411000 125 0 0 0

* Serial0/3/0:0 0 0 0 0 1407000 125 215000 430 0

* Serial0/3/1:0 0 0 0 0 1407000 125 1000 4 0

Is there anyway I can load balance my outbound traffic?

Thank you,

Scott

[Harold Ritter]

Scott,

Are you using a default static route to each of the 4 serial interfaces? If so, could you provide us with a "show ip cef 0.0.0.0" internal.

Thanks,

[network_dude]

Yes, I am using a default route to each...please see ip cef below:

#show ip cef 0.0.0.0 internal

0.0.0.0/32, version 1, epoch 0, receive

Thank you.

[Harold Ritter]

Sorry, the command should have been "show ip cef 0.0.0.0 0.0.0.0 internal".

Thanks,

[network_dude]

Here you go...

#show ip cef 0.0.0.0 0.0.0.0 internal

0.0.0.0/0, version 89, epoch 0, per-destination sharing

0 packets, 0 bytes

via xxx.xxx.xxx.1, 0 dependencies, recursive

traffic share 1

next hop xxx.xxx.xxx.1, Serial0/3/1:0 via xxx.xxx.xxx.1/32

valid adjacency

via xxx.xxx.xxx.2, 0 dependencies, recursive

traffic share 1

next hop xxx.xxx.xxx.2, Serial0/3/0:0 via xxx.xxx.xxx.2/32

valid adjacency

via xxx.xxx.xxx.3, 0 dependencies, recursive

traffic share 1

next hop xxx.xxx.xxx.3, Serial0/2/0:0 via xxx.xxx.xxx.3/32

valid adjacency

via xxx.xxx.xxx.4, 0 dependencies, recursive

traffic share 1

next hop xxx.xxx.xxx.4, Serial0/1/0:0 via xxx.xxx.xxx.4/32

valid adjacency

0 packets, 0 bytes switched through the prefix

tmstats: external 0 packets, 0 bytes

internal 0 packets, 0 bytes

Load distribution: 0 1 2 3 0 1 2 3 0 1 2 3 0 1 2 3 (refcount 1)

Hash OK Interface Address Packets

1 Y Serial0/3/1:0 point2point 0

2 Y Serial0/3/0:0 point2point 0

3 Y Serial0/2/0:0 point2point 0

4 Y Serial0/1/0:0 point2point 0

5 Y Serial0/3/1:0 point2point 0

6 Y Serial0/3/0:0 point2point 0

7 Y Serial0/2/0:0 point2point 0

8 Y Serial0/1/0:0 point2point 0

9 Y Serial0/3/1:0 point2point 0

10 Y Serial0/3/0:0 point2point 0

11 Y Serial0/2/0:0 point2point 0

12 Y Serial0/1/0:0 point2point 0

13 Y Serial0/3/1:0 point2point 0

14 Y Serial0/3/0:0 point2point 0

15 Y Serial0/2/0:0 point2point 0

16 Y Serial0/1/0:0 point2point 0

refcount 3285, covered prefixes:

xxx.xxx.xxx.1/30

xxx.xxx.xxx.2/30

255.255.255.255/32

224.0.0.0/4

xxx.xxx.xxx.xxx/29

xxx.xxx.xxx.xxx/30

xxx.xxx.xxx.xxx/30

0.0.0.0/32

#

[Harold Ritter]

This output shows that per destination (which is really per flow, meaning src and dest IP address are considered) will take place. It is possible that loadsharing will not make use of all interfaces equally if you only have a few flows. The probability of equal loadsharing usually increases when the number of flows is higher.

Hope this helps,

[network_dude]

That makes sense...I'll generate more traffic for further testing.

Thank you for your help!