I am resilience testing an application that is spread across multiple servers.
One thing I will need to do soon is throttle the network traffic for specific interfaces within the test cluster. Specifically, maybe make a connection take twice or three times as long to respond....
I have no experience in this area and would be grateful for any pointers on how to go about it or open source tools that will allow me to do this.
Whatever the solution, it will be run from a shell script on Centos Linux servers.
There are a lot of options for Linux, which all boil down to the Hierarchical Token Bucket or Class Based Queueing in Linux.
Like iptables is, these are kernel drivers not applications, so not precisely friendly. Often a generator script of some sort is used to build a list of commands. I have occasionally used htbinit.