Etherchannel between 2900 and 7200

[cherenkoff]

Hello,

We have Etherchannel trunk set up between Catalyst 2924 switch and 7200 router. The trunk consists of 2 links with source-based forwarding, STP disabled on switch and on the router. For some reason, we can't achieve load-balancing on the trunk, that is, one link is currently used at 100%, another at 20%. We would like to achieve 50/50 utilization. Is it possible to do?

Thanks !

Konstantine

[dnewell24]

How are you deriving your current utilization? Is it source based from the router or switch? Can you post the output?

How many hosts exist in the network? Are they directly connected to the 2924 or is there another L3 device between the host and 7200. Is the 7200 the default gateway? How many subinterfaces exist on the 7200?

Additional information would help or you could try destination-based forwarding.

Ryan

[cherenkoff]

The Etherchannel trunk consists of 0/11 and 0/12 interfaces.Here's the output from "Show interface" command on a switch:

FastEthernet0/11 is up, line protocol is up

Hardware is Fast Ethernet, address is 0001.96bb.51cb (bia 0001.96bb.51cb)

MTU 1500 bytes, BW 100000 Kbit, DLY 100 usec,

reliability 255/255, txload 219/255, rxload 49/255

Encapsulation ARPA, loopback not set

Keepalive not set

Full-duplex, 100Mb/s, 100BaseTX/FX

ARP type: ARPA, ARP Timeout 04:00:00

Last input 2w6d, output 00:00:40, output hang never

Last clearing of "show interface" counters 5w0d

Queueing strategy: fifo

Output queue 0/40, 0 drops; input queue 0/75, 0 drops

5 minute input rate 19376000 bits/sec, 7347 packets/sec

5 minute output rate 86004000 bits/sec, 8 packets/sec

FastEthernet0/12 is up, line protocol is up

Hardware is Fast Ethernet, address is 0001.96bb.51cc (bia 0001.96bb.51cc)

MTU 1500 bytes, BW 100000 Kbit, DLY 100 usec,

reliability 255/255, txload 22/255, rxload 58/255

Encapsulation ARPA, loopback not set

Keepalive not set

Full-duplex, 100Mb/s, 100BaseTX/FX

ARP type: ARPA, ARP Timeout 04:00:00

Last input 00:00:08, output 00:00:01, output hang never

Last clearing of "show interface" counters 5w0d

Queueing strategy: fifo

Output queue 0/40, 0 drops; input queue 0/75, 0 drops

5 minute input rate 22747000 bits/sec, 7947 packets/sec

5 minute output rate 8678000 bits/sec, 2548 packets/sec

As you can see, the output rate on interface 0/11 is 300 times lower than on 0/12 and that is our main problem. We have a web farm here (around 100 servers), which is connected via Layer 4 switch to 2924 and then, via Etherchannel trunk, to 7200 router which is default gateway. We tried to use destination-based forwarding (on a switch), but that didn't change anything. The router AFAIK doesn't have any settings to change the forwarding.

[bhedlund · Accepted solution]

If you look at the input rates you can see the router is evenly loadbalancing the traffic toward the 2924 switch.

Looking at the output rates however (traffic from the switch to the router) we see a different story. The load balancing is very uneven. It would be safe to assume that the imbalacing intelligence (or lack thereof) lies within the switch.

Question: Is this L4 switch connected to the 2924 doing server loadbalancing? Is it hiding the real MAC addresses of the servers from the 7200 router? If so, everything makes perfect sense here. The 2924 sees most of the traffic leaving the switch as coming from one MAC address ( the loadbalancer on the L4 switch ).

In any case, you should never expect to get 50/50 loadbalancing on an etherchannel, escpecially on an older platform such as the 2924 that can only look at MAC addresses for its algorithm.

My suggestion would be to invest $2K in a 3500 series switch and use the command 'port-channel load-balance src-dst-ip' or 'port-channel load-balance src-dst-port'. This should get you about 60/40 in both directions.

Please rate all helpful posts.

Brad

[cherenkoff]

Indeed, the L4 switch was masquerading real MAC addresses. Now it's all clear for me.

Thanks !