MPLS TE Load Balancing

lvpin Lv · ‎06-22-2013

Hello everyone I am a for students from China, here is my problem, my English is not good, this is done by google translate.

Lab Topology。

Lab requirements

With 【R1】 as the source， Ping 【R3】

R1---R2----R3 2 packets

R1 --R4---R5--R3 3 packets

achieve Unequal-cost load balancing

The present situation

After the completion of the relevant configuration

Send data packets R1--R4---R5--R3

Reply data packets R1--R2---R3

Here are my main configuration information

【R1】

interface Tunnel1

ip unnumbered Loopback0

tunnel mode mpls traffic-eng

tunnel destination 30.1.1.1

tunnel mpls traffic-eng autoroute announce

tunnel mpls traffic-eng priority 7 7

tunnel mpls traffic-eng bandwidth 100

tunnel mpls traffic-eng path-option 10 explicit name R1-2-3

tunnel mpls traffic-eng load-share 2

no routing dynamic

!

interface Tunnel2

ip unnumbered Loopback0

tunnel mode mpls traffic-eng

tunnel destination 30.1.1.1

tunnel mpls traffic-eng autoroute announce

tunnel mpls traffic-eng priority 7 7

tunnel mpls traffic-eng bandwidth 100

tunnel mpls traffic-eng path-option 10 explicit name R1-4-5-3

tunnel mpls traffic-eng load-share 3

no routing dynamic

ip explicit-path name R1-2-3 enable

next-address 20.1.1.1 *[R2]

next-address 30.1.1.1 *[R3]

!

ip explicit-path name R1-4-5-3 enable

next-address 40.1.1.1

next-address 50.1.1.1

next-address 30.1.1.1

Data packet analysis

【R1.Ethernet1/0】

SHOW

R1#show ip route 30.1.1.1

Routing entry for 30.1.1.1/32

Known via "ospf 1", distance 110, metric 21, type intra area

Last update from 30.1.1.1 on Tunnel1, 00:07:30 ago

Routing Descriptor Blocks:

30.1.1.1, from 30.1.1.1, 00:07:30 ago, via Tunnel1

Route metric is 21, traffic share count is 2

* 30.1.1.1, from 30.1.1.1, 00:07:30 ago, via Tunnel2

Route metric is 21, traffic share count is 3

R1#show ip cef 30.1.1.1 internal

< 0 > IP midchain out of Tunnel1 676918C0 label 16 TAG adj out of Ethernet1/0, addr 12.1.1.2 6819F3A0

< 1 > IP midchain out of Tunnel2 67691600 label 16 TAG adj out of Ethernet1/1, addr 14.1.1.4 67691A20

< 2 > IP midchain out of Tunnel1 676918C0 label 16 TAG adj out of Ethernet1/0, addr 12.1.1.2 6819F3A0

< 3 > IP midchain out of Tunnel2 67691600 label 16 TAG adj out of Ethernet1/1, addr 14.1.1.4 67691A20

< 4 > IP midchain out of Tunnel1 676918C0 label 16 TAG adj out of Ethernet1/0, addr 12.1.1.2 6819F3A0

< 5 > IP midchain out of Tunnel2 67691600 label 16 TAG adj out of Ethernet1/1, addr 14.1.1.4 67691A20

< 6 > IP midchain out of Tunnel1 676918C0 label 16 TAG adj out of Ethernet1/0, addr 12.1.1.2 6819F3A0

< 7 > IP midchain out of Tunnel2 67691600 label 16 TAG adj out of Ethernet1/1, addr 14.1.1.4 67691A20

< 8 > IP midchain out of Tunnel1 676918C0 label 16 TAG adj out of Ethernet1/0, addr 12.1.1.2 6819F3A0

< 9 > IP midchain out of Tunnel2 67691600 label 16 TAG adj out of Ethernet1/1, addr 14.1.1.4 67691A20

<10 > IP midchain out of Tunnel1 676918C0 label 16 TAG adj out of Ethernet1/0, addr 12.1.1.2 6819F3A0

<11 > IP midchain out of Tunnel2 67691600 label 16 TAG adj out of Ethernet1/1, addr 14.1.1.4 67691A20

<12 > IP midchain out of Tunnel2 67691600 label 16 TAG adj out of Ethernet1/1, addr 14.1.1.4 67691A20

<13 > IP midchain out of Tunnel2 67691600 label 16 TAG adj out of Ethernet1/1, addr 14.1.1.4 67691A20

<14 > IP midchain out of Tunnel2 67691600 label 16 TAG adj out of Ethernet1/1, addr 14.1.1.4 67691A20

R1#show ip route

30.0.0.0/32 is subnetted, 1 subnets

O 30.1.1.1 [110/21] via 30.1.1.1, 00:08:37, Tunnel1

[110/21] via 30.1.1.1, 00:08:37, Tunnel2

I hope get help

Tell me where to errors

Thank you

Here are GNS3 configuration files

Harold Ritter · ‎06-25-2013

This is normal. CEF does "per flow" load balancing by default. So if you are only sending one flow (one source to one destination), only one path will be used.

Regards

lvpin Lv · ‎06-26-2013

Thank you very much Harold Ritter answer, I know

no ip cef then

Lab Phenomenon

Very, very grateful

Harold Ritter · ‎06-26-2013

mpls needs cef to work properly, so "no ip cef" is not an option. You can enable "per packet" load balancing though. This can be achieved by configuring "ip load-sharing per-packet" on the egress interface.

Regards

lvpin Lv · ‎06-27-2013

Oh wow。It's really great。

Thank Harold Ritter answer.

I'll search out all the relevant documents later.

【http://www.cisco.com/en/US/docs/ios/12_0s/feature/guide/pplb.html#wp1015363 】

But in the GNS3, no related commands，just such

R1(config)#hw-module slot 1 ?

shutdown Shutdown a PA

Then I enter an interface ip load-sharing per-packet has no effect

This is regrettable

However, this not a problem of device，But rather my error

When I was on the tunnel interface configuration ip load-sharing per-packet Perfect experimental effect

I had to once again thank the wonderful answer Harold Ritter