Hello, I'm looking to simulate the uneven allocation behavior of TCP for a network performance isolation project. I have a Devstack environment setup with OVN containing two hypervisors, each with two VMs. There are two logical networks which are isolated from each other, where a single VM from each hypervisor is connected to each network. Graphically, it can be represented as follows:
HV1 HV2 ----- ----- | A | -- r1 -- | C | | | | | | B | -- r2 -- | D | ----- ----- where HV1 and HV2 are two hypervisors hosting two VMs each. VMs A and B reside on HV1 while VMs C and D reside on HV2. A and C form one logical network connected by logical router r1 and B and D form the other logical network connected by logical router r2. I'm running a network performance test with iperf across the VMs (with the VMs on HV2 as the iperf servers and the VMs on HV1 as the iperf clients). The test opens 10 TCP connections on one network and a single TCP connection on the other, which I expect to produce a 10:1 bandwidth allocation ratio. However, the test results are indicating a 1:1 allocation ratio. Does anyone have any ideas about what might be causing this and how to solve this problem? Thanks, Brendan
_______________________________________________ discuss mailing list discuss@openvswitch.org http://openvswitch.org/mailman/listinfo/discuss
