on multilink interface ip load sharing per-packet is not available on 7609

Unanswered Question
Mar 10th, 2009
User Badges:

7609(config)# int mul 11

7609(config-if)#ip load-sharing

7609(config-if)#ip load-sharing ?

per-destination Deterministic distribution


7609-2(config-if)#ip load-sharing

------

dCEF is enabled,

is there any hidden command or any software bug, i am using 12.2(33)SRA5 or suggest IOS which supports this

------

the following command is also not available

Globalconfig#hw-module slot <no> per-packet


it is very urgent kindly help me out

  • 1
  • 2
  • 3
  • 4
  • 5
Overall Rating: 3 (1 ratings)
Loading.
pkurdziel Tue, 03/10/2009 - 06:20
User Badges:

ip load-sharing is used under the physical interface.


With a multi link you are already load sharing with the circuits that are bundled.

sgopikrishnan Tue, 03/10/2009 - 06:36
User Badges:

In 7206 the same is working fine on mulilink interface.

In 7609 on the physical interface also shows the same (per-packet option is not seen)

Giuseppe Larosa Tue, 03/10/2009 - 07:37
User Badges:
  • Super Silver, 17500 points or more
  • Hall of Fame,

    Founding Member

Hello Gopikrishnan,

a PPP multilink bundle already perform per PPP frame load balancing over member links on its own, so the practical effect of the command even if supported would be none.



Hope to help

Giuseppe


sgopikrishnan Tue, 03/10/2009 - 10:12
User Badges:

hi Giuseppe,

you mean that in an bundled PPP multilink link it is per packet(by default). But in show commands it does not say.

>>when CEF or dCEF is enabled it is by default per-destination.

-----

You can see the output on the another router 7206 which has 12.2(28).SB8


Multilink interface

>>7206(config)#int mul 11

7206(config-if)#ip load

7206(config-if)#ip load-sharing ?

per-destination Deterministic distribution

per-packet Random distribution

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

Giuseppe Larosa Tue, 03/10/2009 - 10:35
User Badges:
  • Super Silver, 17500 points or more
  • Hall of Fame,

    Founding Member

Hello Gopikrishnan,

actually being each PPP frame sent out one of the member links in a round robin fashion what you get is a fair per packet load balancing.


This something that happens at the MLPPP level code:

for CEF only the L3 interface multilink exists (and this is only one interface).


Then MLPPP feature manages the member links.


So you should already have per-packet load sharing but it is not done by CEF switching process.


you can use

sh ppp multilink


sh int type x/y for each member link to verify that all links are used in a fair way.


Look at the counters and you will see fair use of member links




Hope to help

Giuseppe


sgopikrishnan Tue, 03/10/2009 - 22:35
User Badges:

Hi Giuseppe,

if you have any docs related to this, that would be nice.

Thanks,

Gopikrishnan

sgopikrishnan Wed, 03/11/2009 - 05:04
User Badges:

Hi,


I will check it out and surely revert back for any questions. thank u for ur help. probably by nextweek


Thanks,

Gopikrishnan.s

Actions

This Discussion