Question: Is it possible to take an HP or Dell server with dual 1G Ethernet ports and create a 2G logical EtherChannel (port-channel interface) connection to a Catalyst switch (4948-10GB or 3750) that will give me a total 2G of usable bandwidth? I wouldn't expect to get exactly 2G of bandwidth, but I would like to get close.
I've seen a lot of examples creating EtherChannel connections from servers to Catalyst switches for the purpose of redundancy (in case one port/card/cable fails), but they are still only showing 1G of usable bandwidth.
Maybe I'm misunderstanding the term 'aggregation', but I would expect this to mean that by aggregating 2 - 1G ports, I'd be getting 2G of bandwidth.
Cisco TAC is telling me differently. This is from a case I have openedâ¦
>From the case notes, I understand that you have a port channel with 2
>gigabit Ethernet ports. You stated that you are just getting 1 GB of
>I have to tell you that is perfectly normal and it is the way the ether
>channel works. Your configuration is perfect and everything seems to be ok.
>Let me explain why this is an expected behavior. The ether channel is not a
>feature to increase the throughput of your network or a link between two
>devices. This is a feature that improves the interfaces performance by
>avoiding the traffic to overwhelm one single interface. For example let say
>that you have a port channel of 5 ports this does not mean that you will
>have throughput of 5GB, what this means is that the traffic has 5 options to
>reach an specific device. Which means that you can split 1GB of throughput
>on 5 different interfaces depending of the algorithm you use. You can check
>the algorithm on the following link:
To restate my question(s)â¦
Can I get 2G of usable bandwidth from 2 - 1G ports on my server? Or is the TAC engineer right?
If the TAC engineer is wrong, does anyone have any examples/references to point me in the right direction?