>Problem with the server? switch? Aggregation software? All the above? Maybe not.
>At any rate, I seem to be missing something.. I would hazard a guess that you are misunderstanding aggregation. Any single conversation from _one_ host to _another_ cannot go through two interfaces at once, yielding twice the bandwidth. You can simply load balance multiple conversations for the combined throughput of the aggregate link. If you have n 1gig interfaces, you can only sustain a 1gig conversation with a host, but you can have n 1gig conversations with n hosts (in reality you wont saturate a link completely off one conversation etc...) I am guessing your testing methodology is flawed. If your server is setup with a link aggregate, setup multiple clients and then measure the combined throughput the aggregate provides while communicating with all the clients. jlc
