VXLAN Performance on vSphere 5.1

The VMware vSphere I/O performance team has published a paper that shows VXLAN performance on vSphere 5.1. Virtual extensible LAN (VXLAN) is a network encapsulation mechanism that enables virtual machines to be deployed on any physical host, regardless of the host’s network configuration.

The paper shows how VXLAN is competitive in terms of performance when compared to a configuration without VXLAN enabled. The paper describes the test results for three experiments: throughput for large and small message sizes, CPU utilization for large and small message sizes, and throughput and CPU utilization for 16 virtual machines with various message sizes. Results show that a virtual machine configured with VXLAN achieved similar networking performance to a virtual machine without VXLAN configured, both in terms of throughput and CPU cost. Additionally, vSphere 5.1 scales well as more virtual machines are added to the VXLAN network. Read the full paper here.


3 comments have been added so far

  1. Very interesting results. I like that VMware is showing these kinds of results.
    I do have a few questions?

    1) Why was RSS not enabled on the test where VXLANs were not enabled?
    2) What was the impact on latency for these tests? Since VXLAN requires the CPU to be engaged to due checksum work, I would assume that there might be an increase in both network latency and VM CPU latency.

    1. Eli,

      We enabled RSS for making sure that VXLAN traffic can use multiple hardware queues. Without RSS, VXLAN traffic will go to a single queue which will hurt performance. For the default case, multiple VMs will use multiple queues irrespective of whether RSS is enabled or not. Also there is no way to configure a single VM to use RSS on the physical NIC.

      We did some latency experiments and didn’t observe a significant impact of using VXLAN.


Leave a Reply

Your email address will not be published. Required fields are marked *