IP Networking

Hello, I am trying to get clarity on a few things and am looking for some info.

In every article I have read about link aggregation and lacp, it can be used combine physical links to create 1 logical link for increased bandwidth.

But what it doesn't say is if this is limited by source/dst. IE 1 destination can only reach the capacity of one of the physical links, or if in fact it is a true 4GB connection. I have looked into several hashing methods, and linux bonding seems to support round robin which would do what I am looking for. However, VMware/ESX does not have this type of bonding, just IP/Mac/Port hasing.

I have had these results trying to setup a 4GB connection from ESX<->switch, and Linux/NFS<->switch over 4 1gb nics. In the end results I am trying to get 320/MB disk transfer between the vmware server and the Linux/NFS/Disk array I have setup. Is LACP the wrong choice maybe? Or has anyone been able to set this up and if so what type of hardware?

If you have a source DTE with 4 1gb/s interfaces trunked into a duplex capable 10g switch then you should easily attain your desired throughput. Whether the end host system can receive 320/mbs throughput depends on it's configuration. It's noted that lacp is far from perfect without factoring in the latencies from NFS and/or VM translation. Also, your backend potential throughput is faster than a lot of local
disk buses.