Hello,
We want to implement a qos policy to alocate a maximum amount of bandwidth to a host-to-host ftp transfer, during times of congestion only. We are using 3550 mls switches.
we are marking this traffic with dscp on the server port and then mapping the dscp to an egress queue. The egress quese has a set bandwidth allocation.
When testing to varify the config, we cannot create congestion on the interface, to invoke the hardware queues. We have been using ftp sessions and a very basic ip load generator. ( cannot find anything else that is free!)
the traffic is all TCP based and gets very close to using the available bandwidth, but will not go all the way. Is it because of the nature of TCP and it's windowing mechanism?
any suggections please?