We are building a test client to do some load/failover testing for Apache Qpid.
My admin counterparts told me that I should randomize the bytes in the message payload. Their reasoning for this is that there is some type of caching going on that could impact the validity of the tests.
Before I go to all the trouble of randomizing all the payload bytes(which itself will have an impact on the test results), Is this true? Is there caching that will occur at the interface level that I should be aware of? Do you think it could significantly impact my test results? Consider that beyond the payload of the message are other fields that will change(Message Id, Destinations, meta-data, etc.)
I think your colleague is mixing up TCP and HTTP. The only caching that I know goes on at the interface level is ARP caching and some linux distros offer network statistic caching, which you can cut off via a parameter.