C3750X configure etherchannel with netapp

[khuyenhoang]

Hello Cisco experts,

I have configured Cisco C3750X-48TS etherchannel with netapp FAS2240 configured  static multi_mode .

interface Port-channel1

description FAS2240

flowcontrol receive on

spanning-tree guard loop

!

interface GigabitEthernet1/0/37

flowcontrol receive on

no cdp enable

spanning-tree guard loop

channel-group 1 mode on

!

interface GigabitEthernet1/0/38

flowcontrol receive on

no cdp enable

spanning-tree guard loop

channel-group 1 mode on

After finish i make a test by transfering bulk data between host to Netapp 2240

and using command :

>ifgrp stat ifgrp01 on netapp. Here is a result from Netapp 2240.

fas2240a*> ifgrp stat ifgro   p01

Interface group(trunk) ifgrp01

        e0a                 e0c        

  Pkts In   Pkts Out  Pkts In   Pkts Out

  1458m     738m      1574m     537m   

  37        0         24257     19488  

  26        0         29780     24335  

  21        0         27625     21592  

  26        0         22600     18088  

  36        0         26997     21787  

  29        0         25093     20106  

  23        0         24649     20462  

  20        0         27884     22852  

  33        0         23202     18472  

  24        0         27121     21215  

  23        0         23637     18001  

  21        0         23001     17625  

  24        0         22984     17725  

  32        0         22341     17352  

  21        0         21960     16922  

  23        0         20662     16444  

  32        0         18180     13994  

  25        0         15177     11262  

  27        0         13175     9078   

  18        0         13819     9482   

  31        0         14433     9975   

the interface e0a seem doesn't work only e0c has data load on it.

Could you please advice me how to correct my trouble!

Thanks so much!

[Jon Marshall]

Different switches support different load balancing methods eg. src/dst mac, src/dst IP, src/dst port but the one thing common to etherchannel load balancing is that a single conversation between two devices will always be sent down one specific link. This is what you are seeing ie. at one end you have the netapp device and the other the host. This is a single conversation so it will always use the same physical port in the etherchannel group. 

Now lets say you had another host and started a conversation between that host and the netapp device. This is now seen as a different conversation and may well be allocated to the other link.

So what you are seeing is normal behaviour. If you actually needed more than 1Gbps throughput between a specific host and the the netapp device etherchannel is not the solution ie. you would need faster interfaces in the devices.

Jon